-
AI Sales Training ROI: How to Calculate It (Snowflake Example)
AI sales training ROI is the return a sales team gets from AI-powered practice and certification, measured against what the program costs. It comes down to a few inputs that most enablement teams already have or can collect in a week. These include how many hours managers spend grading practice, how long certification takes, and how fast new reps ramp. This guide walks through the cost inputs, the outcome inputs, a simple formula, a worked example, and how to report the result. It uses the published Yoodli Snowflake case study as a real-world reference point.
What AI Sales Training ROI Measures
ROI compares the monetary value a program produces with what it costs. For AI sales training, the biggest share of value usually comes from manager time, because grading and mock calls add up fast across a large team. Faster certification and faster ramp for new hires add to it. Skill gains and better win rates sit on top of those, and they take longer to show.
Start from the business result and work backward. Kirkpatrick Partners, which maintains the Kirkpatrick Model, says that “Whenever applying the Kirkpatrick Model, it is critical to start with Level 4: Results.” In practice, that means agreeing with the sales leader on the outcome the program should move before you build a single roleplay. Yoodli’s guide to the Kirkpatrick Model covers all four levels if you want a refresher.
The Cost Inputs: What Your Current Process Costs
The first half of the ROI model is a baseline. Measure what the current training process costs in time before the AI program starts, because every benefit you report later is a change against that baseline. Three inputs matter most.
Manager grading time
Count the hours frontline managers spend reviewing pitch recordings, scoring roleplay submissions, and sitting in on mock calls. Ask a sample of managers to estimate hours per quarter, or pull it from calendar data for a recent certification cycle. Multiply hours by the number of managers to get a quarterly total.
Certification time
Track how long it takes to get a cohort certified on a new pitch, product, or process. Include rep hours spent on live attempts and scheduling, plus the number of attempts it takes to pass. Long certification cycles delay launches, so the calendar time matters as much as the hours.
Ramp time
Record how many weeks it takes a new hire to reach a defined productivity mark, such as first closed deal or a set share of quota. Yoodli’s post on whether AI roleplay can reduce sales rep ramp time covers which practice habits tend to move this number.
Turning hours into dollars
Use a fully loaded hourly rate for each role, and get it from finance if you can. A wage figure alone understates the cost of an hour. The U.S. Bureau of Labor Statistics Employer Costs for Employee Compensation release for June 2026 reports that wages and salaries accounted for 70.0 percent of private industry employer compensation costs. Benefits made up the remaining 30.0 percent.
The Program Costs
The other side of the ledger is what the AI sales training program itself costs. The ROI Institute’s ROI Methodology guidance says to “Fully load all costs of a solution, project, or program when analyzing ROI.” For an AI roleplay program, that usually means three lines:
- The platform cost for the period you are measuring.
- Enablement time to build scenarios, write rubrics, and run the program.
- Rep time spent practicing, valued at the reps’ loaded hourly rate.
Leaving rep practice time out makes the ROI look better than it is. Sales leaders tend to notice, so include it.
The Outcome Inputs: What the Program Changes
Outcome inputs are the changes you measure after launch, converted into dollars where you can. Keep each one tied to a baseline from the cost inputs above.
- Manager hours returned is baseline grading hours minus grading hours after AI scoring, times the manager loaded rate.
- Certification time saved is baseline rep hours per certification minus new rep hours, times the number of reps and the rep loaded rate.
- Faster ramp is weeks saved per new hire, times the weekly value of a fully ramped rep, times an attribution share.
- Skill improvement is the change in rubric scores across attempts and cohorts. Report it alongside the dollar figures, but don’t convert it to money.
The attribution share matters. Ramp speed depends on territory, hiring quality, and product changes as well as training. The ROI Institute’s methodology includes a step to isolate the program’s effects from other influences, so apply a conservative share instead of claiming the whole improvement.
The AI Sales Training ROI Formula
The ROI Institute gives two formulas. The benefit-cost ratio (BCR) divides program benefits by program costs. ROI as a percentage divides net program benefits by program costs and multiplies by 100.
- BCR = program benefits ÷ program costs
- ROI (%) = (program benefits minus program costs) ÷ program costs × 100
A BCR of 2:1 means every $1 spent returns $2, which is an ROI of 100%. Use the same time period for benefits and costs. A quarter works well, because manager grading hours and certification cycles usually run on a quarterly rhythm. Yoodli’s guide to the Phillips ROI model for training goes deeper on the method.
Worked Example: Calculating AI Sales Training ROI for One Quarter
The numbers below are illustrative assumptions for a mid-sized sales org. Swap in your own baseline and rates.
The team has 60 frontline managers and 250 reps, and it hires 20 new reps per quarter. Managers spend 6 hours each per quarter grading practice, which is 360 hours. With AI roleplays scoring attempts against the team’s rubric, managers spend 1 hour each reviewing flagged attempts, so 300 hours come back. Reps used to spend 2 hours each on live certification attempts and scheduling, and that drops to 1 hour. New hires reach their productivity mark 1 week sooner.
Line Calculation Quarterly value Manager hours returned 300 hours × $120 loaded rate $36,000 Certification time saved 250 reps × 1 hour × $60 loaded rate $15,000 Faster ramp 20 hires × 1 week × $5,000 weekly gross profit × 50% attribution $50,000 Total benefits $101,000 Platform cost Quarterly share of the license $30,000 Enablement build and admin time 80 hours × $90 loaded rate $7,200 Rep practice time 250 reps × 1.5 hours × $60 loaded rate $22,500 Total costs $59,700 Net program benefits are $101,000 minus $59,700, which is $41,300. The BCR is about 1.69:1, and the ROI is about 69% for the quarter. Build costs often fall in later quarters as scenarios get reused, so the ratio can improve over time.
Label the manager hours line as capacity returned to coaching. The managers are still on the team, and the value comes from spending those hours on deal reviews, call coaching, and one-on-ones instead of scoring recordings.
A Real Example: Snowflake’s Manager Grading Time
The Snowflake case study shows the manager-grading input in practice. Before the program, 162 district managers were spending an estimated average of 7.5 hours each per quarter grading pitch and roleplay submissions, which equated to 1,215 hours. Snowflake moved that scoring to Yoodli AI roleplays, which gave reps rubric-based feedback in their native language.
According to the case study, the program eliminated about 1,215 hours of manual grading per quarter and saved an estimated $170K quarterly. The hours follow the same baseline method as the worked example above, which is the number of managers times hours per manager.
Participation is the input that makes those numbers credible. Snowflake certified nearly 3,000 sellers and managers, 94% of the target audience, in just a few weeks. Yoodli’s tool training page shows how teams run certifications like that for platform rollouts. High completion matters for ROI, because grading hours only come back when reps actually certify through the new process.
How to Report AI Sales Training ROI
Report early and on a schedule the sales leader agreed to. Leading inputs like participation, certification completion, and manager hours returned can be counted in the first quarter. Ramp and win-rate effects take longer, because enough reps and deals need to move through the funnel first.
A one-page quarterly report usually covers four things:
- The business outcome the program targets, in the sales leader’s own words.
- Participation and certification completion by team.
- The ROI calculation, with each assumption listed, including the loaded rates and the attribution share.
- Skill progress against the rubric, with the criteria that improved and the ones that still need coaching.
Yoodli Measure shows progress by learner, team, and program against your rubrics, which covers the participation and skill sections of that report. Pair it with certification data from Yoodli Roleplays and your own baseline numbers for the dollar lines. Yoodli’s breakdown of the sales enablement metrics leadership actually wants to see helps decide what goes on the page.
Show where the returned manager hours went. A short note from managers on what they did with the time makes the capacity line concrete. Yoodli’s post on how managers can use AI coaching data in 1:1s has practical ideas for spending it well. For onboarding programs, Yoodli’s onboarding and certification page shows how teams structure the cohorts that feed the ramp line.
Yoodli is the AI Platform for Human Skills, built to help people get measurably better at their jobs. To scope an ROI model around the outcome your team needs to prove, talk to the Yoodli team.
FAQ
How do you calculate AI sales training ROI?
Calculate AI sales training ROI by subtracting program costs from program benefits, dividing by program costs, and multiplying by 100. Benefits usually include manager grading hours returned, certification time saved, and faster ramp, each converted at a loaded hourly rate. Costs include the platform, enablement build time, and rep practice time, all for the same period.
What costs should go into an AI sales training ROI model?
Include every cost of running the program for the period you measure. That means the platform cost, enablement time spent building scenarios and rubrics, and rep time spent practicing at a loaded hourly rate. On the baseline side, record current manager grading hours, certification time, and ramp time before launch so you can measure the change.
How long does it take to see ROI from AI sales training?
Manager hours returned and certification completion can usually be counted within the first quarter, if you set a baseline before launch. Skill improvement shows up across repeated attempts inside a program. Ramp and win-rate effects often take several quarters, because enough new hires and deals need to move through the pipeline before the data means anything.
What ROI did Snowflake see from AI roleplays?
The Snowflake case study reports that Yoodli AI roleplays eliminated about 1,215 hours of manual grading per quarter and saved an estimated $170K quarterly. Before the program, 162 district managers spent an estimated average of 7.5 hours each per quarter grading submissions. Snowflake also certified nearly 3,000 sellers and managers, 94% of its target audience.
-
Contact Center Training: How to Get Agents Ready for Real Customer Calls
Contact center training gets agents ready for real customer calls in two parts. Agents learn the products, systems and policies, then practice the conversations out loud until they meet the standard your QA team uses on live calls. The practice part decides whether the training holds up. An agent who has read the refund policy can still freeze the first time a customer who already tried the app twice asks for a supervisor.
This guide is for contact center, CX and support training leaders who own the whole program, from the first class of new hires to the tenured agents on the floor. For a week-by-week onboarding plan, our guide to call center agent training covers a new hire class in detail. Here the focus is the program around onboarding: what to train, where contact center roleplay fits, how to certify readiness and how to keep skills current after go-live.
Why the calls that reach agents keep getting harder
Customers try self-service first, so the contacts that reach a person are usually the ones self-service couldn’t solve. That moves training toward judgment, empathy and complex troubleshooting.
Gartner found that only 14% of customer service and support issues are fully resolved in self-service, based on a survey of 5,728 customers. Even for issues customers describe as “very simple,” only 36% resolve fully in self-service. A later Gartner survey of 265 customer service and support leaders found that “the remaining assisted service interactions are more complex” now that many simple cases are resolved in self-service.
Contact center leaders report the same thing. Calabrio’s 2025 State of the Contact Center report surveyed 437 contact center managers in 13 countries. Almost all of them use AI in some form, and most say customer conversations have become more challenging. Training and coaching have been slower to catch up.
Finding Share Use AI in contact center operations at some level 98% Say customer conversations are more challenging 61% Say emotional intelligence training is not available to agents 64% Don’t provide ongoing coaching for AI-driven workflows 59% Nearly every contact center uses AI, and most leaders say the conversations left for agents are harder. Source: Calabrio, 2025 State of the Contact Center
Calabrio also found that leaders see empathy as the skill agents lack most. Empathy is hard to learn from a slide deck. Agents build it by handling upset customers, hearing feedback on what they said and trying the same call again.
What contact center training should cover
A complete contact center training program covers four areas: knowledge, systems, conversation skills and compliance. Agents can learn the first two by reading and doing. The last two have to be practiced out loud, because they break down under pressure.
- Knowledge: Agents learn the products, plans, pricing, policies and answers to your most common questions.
- Systems: Agents learn the CRM, ticketing tool, knowledge base and phone or chat platform, including how to find an answer while the customer waits.
- Conversation skills: Agents practice opening the call, asking the right questions, explaining a charge in plain language, de-escalating and closing with a clear next step.
- Compliance: Agents practice identity verification, required disclosures and anything regulated in your industry. Our guide to call center compliance training covers how to certify these behaviors.
The program also has to run past onboarding. Many teams run four kinds of training through the year. There’s new hire onboarding, then change training when a product, policy or system launches. QA findings trigger targeted practice for the agents who need it. Tenured agents get advanced skills work for escalations, retention calls and technical cases.
Self-service adds one more conversation skill to the list. Gartner found that 60% of customer service agents fail to promote self-service options, and agent promotion is associated with a doubling of the customers likely to adopt self-service next time. Showing a customer where to solve the issue on their own is something agents say on the call, so it belongs in practice too.
How to build a contact center training program
Build the program backward from the contacts agents actually handle and the standard QA uses to score them. The same six steps work for a new hire class and for a single policy change.
Step What it involves 1. Map contact drivers List the top reasons customers call or chat 2. Set the standard Use your QA form as the training rubric 3. Write scenarios One per driver, at three difficulty levels 4. Practice out loud Agents run each scenario until they pass 5. Certify Scored calls on core contact types before the queue 6. Feed QA back in Turn gaps from live QA into new practice Strong programs start from real contact drivers and end with live QA findings feeding new practice.
Map your contact drivers
Pull the most common contact reasons from your ticketing system, call recordings and chat logs. Rank them by volume and by risk. A billing dispute might be your most frequent call, while a fraud report might be rare but costly to get wrong. Both belong on the list.
Use your QA form as the rubric
If live calls are scored on verification, accuracy, empathy, resolution and closing, score practice the same way. Agents then hear one standard from their first day, and trainers and QA analysts work from the same definition of a good call. Our guide to call center quality assurance walks through building that scorecard.
Write scenarios at three difficulty levels
For each contact driver, write a calm version, a frustrated version and a version where the customer already tried self-service and wants a supervisor. That third version matters more every year, given the Gartner numbers above. Each scenario defines the customer’s goal and mood, what the agent must verify and what a good resolution looks like in your QA terms.
How contact center roleplay prepares agents for real calls
Contact center roleplay gives agents a realistic customer to talk to before a real one is on the line. With AI roleplay for contact center training, every agent can run the same call as many times as they need and get scored on your rubric after each attempt.
Mock calls with a trainer still have a place, especially for a new class’s first few calls. The limit is time. A trainer can only listen to so many calls in a shift, so agents get a handful of repetitions on the calls they’ll take hundreds of times. AI practice lets agents take the repetitions on their own schedule, and trainers spend their time on the attempts that need a human ear.
Here’s how a training leader might set up a billing dispute in Yoodli. In Yoodli Personas, you set the customer’s voice, mood and context: a customer who was charged twice, already checked the help center and is ready to cancel. In Yoodli Roleplays, you attach your QA rubric. The rubric might score whether the agent acknowledged the earlier attempt, verified identity and explained the charge in plain language. It can also check that the agent fixed the problem or set a clear next step, then showed the customer where to check charges next time.
Agents take the call, the AI customer pushes back the way real customers do, and each attempt comes back scored with feedback on what to change. Training leads build the same scenario at all three difficulty levels and assign them by class or by agent. Because the scoring uses your rubric, a practice score means the same thing as a QA score.
The emotional calls deserve the most repetitions. Our guide to de-escalation training for customer service covers the techniques to build into those scenarios. For global teams, Yoodli supports 25+ languages, so each site can practice in the language its customers speak.
Certify agents before they join the queue
Certification is the readiness gate. Each agent passes a scored roleplay on every core contact type before taking live contacts alone. Readiness then rests on a score against the QA standard instead of a trainer’s impression.
RingCentral shows what this looks like at scale. Its managers used to observe and grade each mock call by hand, which could take up to 9 hours per new agent. After moving mock call certification into Yoodli, RingCentral cut support call certification time by 90%, from 9 hours to 1 hour per agent. Trainees got immediate feedback and could retry until they met the quality standard. Training leads and supervisors saved up to 7 hours per certification, which freed time for coaching.
Certification also works after onboarding. When a policy changes, add the new scenario and have every agent certify on it before the change goes live. Recertify on high-risk contact types, such as compliance calls, on a regular schedule. Teams that run onboarding and certification on one platform keep every class on the same standard, whatever the site or shift.
Keep tenured agents sharp with QA and calibration
After go-live, contact center training runs on QA data. Each review should point an agent to one skill and one practice scenario on that skill, so coaching turns into repetitions before the next review.
Look for patterns across the floor too. If many agents miss the same verification step, write a new scenario or a clearer policy explanation and assign it to the whole team.
All of this depends on QA scores that agree. If two reviewers score the same call differently, agents get mixed signals and practice targets the wrong skill. Regular call calibration sessions keep reviewers aligned, and the rubric they agree on becomes the rubric your practice scenarios use. Yoodli Measure shows practice progress against those rubrics by learner, team and program, so training leads can see which skills are improving before the next QA cycle.
How to choose an AI roleplay tool for contact center training
Pick a tool that practices your calls against your standard. Use this checklist in demos and ask each vendor to build one of your real contact drivers live.
- Look for customer personas that sound real, show emotion and push back when the agent gets it wrong.
- Check that it scores against your own QA rubric, with feedback tied to specific moments in the call.
- Ask how fast it builds scenarios from your knowledge base, scripts and policies.
- Confirm it certifies agents against a clear pass standard and reports readiness by agent and team.
- Confirm it supports every language your sites work in.
- Check that it embeds in your LMS, so practice sits inside the learning path agents already use. Yoodli embeds through SCORM, LTI or a web embed.
- Ask for enterprise security, including SOC 2 Type 2, GDPR and EU AI Act compliance.
Start a pilot on your hardest contact driver, the one that produces the most escalations or repeat contacts. You’ll learn more from that one scenario than from ten easy ones. You can see how support teams set up these programs on the Yoodli page for customer support training.
How to measure training results
Measure readiness before go-live and performance after it. Practice data is the early signal, and live metrics tell you whether it held up once agents faced real customers.
Before agents join the queue, track practice scores by contact type, attempts needed to pass and certification pass rates by class. After go-live, track QA scores, first contact resolution, customer satisfaction, handle time and escalation rate. Compare classes and sites against each other. When one class certifies faster but scores lower on live QA, your scenarios are probably too easy, so raise the difficulty.
FAQ
What is contact center training?
Contact center training is the program that prepares agents to handle customer calls, chats and emails to a set standard. It covers product knowledge, systems, conversation skills and compliance, and it runs from new hire onboarding through ongoing coaching. Strong programs have agents practice real contact types out loud and certify before they take live contacts.
How is a contact center different from a call center for training?
A call center handles mostly phone calls, while a contact center supports every channel, including phone, chat, email and messaging. The core skills overlap. Contact center programs add channel skills, such as writing clear chat replies and handling several chats at once. They also need practice scenarios for each channel agents work in.
What is contact center roleplay?
Contact center roleplay is a practice call where an agent handles a realistic customer scenario before facing real customers. A trainer, a peer or an AI customer plays the caller with a set issue and mood. The agent is scored on the same rubric QA uses, and AI roleplays let every agent repeat the call until they pass.
How do you keep tenured agents’ skills sharp?
Use QA reviews to pick one skill per agent, then assign a short practice scenario on it. Add new scenarios whenever a product, policy or system changes, and have every agent run them before the change goes live. Recertify on high-risk contact types, such as compliance calls, on a regular schedule.
Get agents ready before the first hard call
The hardest call an agent takes in their first month should never be the first time they’ve heard it. Map your contact drivers, score practice on your QA form and certify before the queue. Then let live QA tell you which scenario to write next. Agents who have already handled the angry customer, the double charge and the supervisor request in practice handle them better on the floor.
-
How Do You Identify Sales Coaching Gaps Across an Entire Team?
To identify sales coaching gaps across an entire team, define the skills that matter and assess sellers against consistent criteria. Then look for patterns in the results and prioritize the gaps that affect the most sellers or carry the most business risk. Use several sources of evidence, including manager observations, customer conversations, structured assessments and AI roleplay performance.
The goal is to separate isolated seller weaknesses from systemic issues with messaging, methodology adoption, onboarding or coaching consistency. A team-wide skills analysis helps sales enablement leaders decide where to focus training time and which sellers need extra support.
What Are Sales Coaching Gaps?
Sales coaching gaps are differences between the selling behaviors an organization expects and the behaviors its sellers consistently demonstrate.
For example, a sales organization may expect every account executive to uncover measurable business impact during discovery. An assessment might then reveal that many sellers ask surface-level questions and never explore the financial or operational consequences of the buyer’s problem. That difference is a coaching gap.
Coaching gaps can exist at several levels:
- Individual gaps: A specific rep struggles with an important skill.
- Team gaps: A group of sellers consistently struggles with the same behavior.
- Organizational gaps: The issue appears across regions, managers or business units.
The distinction matters because the right response differs. An individual gap may call for targeted coaching. An organizational gap may call for changes to training, messaging or how the sales methodology is reinforced. The same logic applies outside sales, and our guide to employee skills gap analysis covers it for L&D teams running company-wide programs.
Why Team-Wide Coaching Gap Analysis Matters
Sales leaders often know which reps are missing quota. Quota attainment alone doesn’t explain why performance differs, though. A seller may miss quota because of weak discovery, poor territory conditions, thin pipeline or product-market challenges.
Coaching gap analysis focuses on the behaviors you can observe and develop. Coaching is worth that focus. Matthew Dixon and Brent Adamson made this case in Harvard Business Review. They reported that Sales Executive Council research shows “no other productivity investment comes close to coaching in improving reps’ performance.”
Sales leaders still lean on coaching today. A Gartner survey of 234 heads of sales and senior sales leaders asked which levers chief sales officers (CSOs) rely on most. 57% of CSOs named some combination of manager coaching, new or improved tools, and upskilling or training among their top three. The same research found that organizations with a data-driven understanding of the seller actions that produce results are 2.5x more likely to be among the top sales organizations. A team-wide view of coaching gaps gives you that kind of understanding.
Gap analysis helps organizations:
- Identify recurring skill weaknesses
- Prioritize enablement investments
- Improve coaching consistency
- Target onboarding improvements
- Reinforce sales methodologies
- Monitor readiness across teams
- Rely less on anecdotal feedback
The Kirkpatrick Model is useful here because it separates demonstrated learning and behavior change from downstream business outcomes. A coaching gap analysis lives mostly at the behavior level, which is where coaching can act.
Seven Steps to Identify Sales Coaching Gaps Across a Team
Step 1: Define the Skills That Matter
Start with a competency framework that matches your sales process. These are common skills and the behavior that shows each one:
Skill Observable behavior Discovery Asks relevant questions and investigates buyer needs Active listening Responds to what the buyer said instead of a rigid script Business-impact exploration Clarifies operational or financial consequences Value articulation Connects capabilities to buyer priorities Objection handling Clarifies concerns before responding Executive communication Communicates concise, relevant business value Next-step management Establishes clear mutual actions A sales competency framework ties each skill to a behavior a reviewer can observe and score.
Avoid measuring too many skills at once. Start with the behaviors most closely tied to your organization’s current priorities.
Step 2: Establish Consistent Evaluation Criteria
Define what strong, developing and weak performance look like. For example, a discovery rubric might use a five-point scale:
- 1: The seller doesn’t explore the buyer’s underlying problem.
- 3: The seller identifies the problem but only partly explores its consequences.
- 5: The seller clarifies the problem, explores business impact and confirms understanding.
The exact scale matters less than applying it the same way every time. Managers and reviewers should calibrate their scoring on common examples. Our guide to building a sales call scorecard walks through how to write criteria reviewers can agree on.
Step 3: Gather Evidence From Multiple Sources
Useful sources include:
- Structured roleplay assessments
- Manager coaching notes
- Real customer call observations
- Sales certification results
- CRM qualification quality
- Relevant sales performance trends
Each source has limits. Real calls give field context, but their difficulty varies widely. AI roleplays give a more controlled setting, though they can’t perfectly reproduce every customer interaction. CRM data may reveal qualification gaps, but it can’t show conversation quality on its own. Combining sources gives you a more complete picture. Teams that already record calls can also turn them into practice with post-call coaching from real calls.
Step 4: Identify Patterns Across Sellers
Aggregate the results by skill. For each skill, look at two numbers: the average score and the share of sellers below your readiness standard. The skill with a low average and a large share of sellers below standard is usually your biggest team-wide gap.
Before acting on it, review how the assessments were run and whether the sample is representative. A gap that shows up only in one week of calls, or only with one reviewer, may say more about the assessment than about the team.
Step 5: Segment the Results
Overall averages can hide important differences. Break the results down by groups such as role, tenure, region, product line, sales segment, manager and onboarding cohort.
For example, new hires may struggle with product positioning while experienced enterprise sellers struggle with executive-level value conversations. Those groups shouldn’t automatically get the same intervention. Be careful with small groups, and with differences caused by unequal assessment conditions.
Step 6: Prioritize the Most Important Gaps
Not every gap needs an immediate training program. Use three criteria:
- Prevalence: How many sellers are affected?
- Business relevance: How important is the skill to the sales motion?
- Coachability: Can practice and feedback improve the behavior?
Those three criteria turn into a simple priority framework:
Priority Characteristics High Widespread, important and coachable Medium Important but limited to a specific group Low Infrequent or not business-critical right now Sales coaching gaps that are widespread, important and coachable go to the top of the list.
Prioritizing keeps enablement teams from building training for every weakness they observe.
Step 7: Reassess After Coaching
Gap analysis is only useful if it leads to action. After a coaching intervention, reassess the same behaviors with comparable criteria and answer five questions:
- Did the targeted skill improve?
- Did the improvement hold across multiple attempts?
- Did sellers apply the skill in customer conversations?
- Are some groups still struggling?
- Does the training program need adjustment?
Those answers feed the next round of assessment, so the process keeps improving.
How AI Roleplays Help Identify Coaching Gaps
AI roleplays give you a repeatable setting for evaluating conversation skills. With Yoodli AI roleplays, organizations build scenarios that match their sales methodology and buyer situations, and Yoodli grades each attempt against the organization’s own rubric.
For example, every new account executive could complete a discovery scenario with an unclear buyer problem, multiple stakeholders, a possible budget constraint and a buyer who gives incomplete information. Because every seller faces the same scenario and the same criteria, patterns in how sellers handle it are easier to spot. Yoodli Measure shows progress by learner, team and program against those rubrics.
Clari used this approach with its customer success managers. The team defined five conversation goals with a top-performing CSM, and CSMs practiced against them in Yoodli. The per-goal scores showed which skill was already strong and where coaching had the most room to help. Clari saw a 36% average improvement across five core conversation skills.
Automated evaluations still need review and calibration. Treat AI roleplay scores as strong development signals and pair them with field observations before you draw conclusions about live performance.
Example: Finding a Team-Wide Objection Handling Gap
Consider a hypothetical 60-person sales team. The enablement leader suspects that sellers are discounting too quickly. The team evaluates a representative sample of pricing objection conversations and runs a structured roleplay assessment.
The findings show:
- Most sellers acknowledge the pricing concern.
- Many respond before clarifying the underlying issue.
- Several introduce discounts before exploring value.
- Stronger performers ask whether the objection is about affordability, budget timing or perceived return.
The team defines one target behavior: ask a relevant clarifying question before discussing a pricing concession.
Enablement builds targeted practice scenarios around that behavior. Managers reinforce it during deal coaching, and the organization reassesses performance after several practice cycles. This approach fixes a specific behavior instead of assigning generic objection handling training. For more techniques, see our sales objection handling guide.
How to Tell Whether a Gap Is Individual or Systemic
One diagnostic question does most of the work: does the same problem appear across multiple sellers, managers or cohorts?
If one seller struggles, individual coaching may be the answer. If an entire onboarding cohort struggles, the training program may need revision. If experienced sellers across multiple regions struggle with the same message, the organization may need clearer positioning or updated enablement materials.
Look for patterns before assuming that every performance issue is a rep-level problem. Manager time is limited. A Gartner survey of more than 300 sales managers found that 58% of sales managers struggle to accomplish all the tasks they have been given in the time allotted. Fixing a systemic gap once, at the program level, saves every manager on the team from coaching the same issue rep by rep.
What Metrics Should Sales Enablement Teams Track?
A practical coaching gap dashboard might include these metrics:
Metric Why it matters Average skill score Indicates general proficiency Share below readiness standard Shows the size of the development need Skill improvement over time Measures progress Repeat practice attempts Shows reinforcement activity Gap distribution by cohort Reveals where support is needed Manager evaluation consistency Supports fair assessment Field behavior observations Validates transfer to real conversations A coaching gap dashboard tracks skill scores, readiness and field behavior together, since averages alone can hide a group that isn’t ready.
Avoid relying on averages alone. A team with a strong average score may still have a sizable group of sellers who aren’t ready for critical conversations. For more on proving that coaching works, see our guide on how to measure sales coaching effectiveness.
How to Turn Gap Analysis Into Personalized Coaching
Once you identify a gap, assign targeted practice instead of generic training. Here is how one gap turns into a practice plan:
- Observed gap: Sellers don’t explore business impact.
- Target behavior: Ask follow-up questions about the consequences of the current problem.
- Practice scenario: A buyer describes a process inefficiency but at first plays down its importance.
- Evaluation criteria: The seller uncovers the business consequence and confirms its relevance.
- Follow-up: Reassess the behavior and check whether it shows up in real calls.
Our guide on how to scale sales coaching covers how to run this across a large team without adding hours to every manager’s week.
Common Mistakes in Sales Coaching Gap Analysis
- Relying only on quota attainment. Results don’t always explain skill gaps.
- Using inconsistent scoring criteria. Comparisons become unreliable when reviewers use different standards.
- Treating one conversation as definitive. Look for repeated evidence.
- Ignoring role differences. Different sales roles need different capabilities.
- Assuming AI scores are perfectly accurate. Calibrate evaluations and use human judgment.
- Focusing only on the lowest scores. Consider business relevance and how many sellers are affected.
- Failing to act on findings. A dashboard improves skills only when it shapes development.
Build a Continuous Sales Coaching Improvement Process
Team-wide coaching gap analysis works best as an ongoing cycle. Define the skills, assess performance, identify patterns, prioritize gaps, assign practice and measure improvement, then start again with the next set of priorities.
Yoodli supports this cycle with realistic AI roleplays, personalized feedback and visibility into practice performance. See how it fits your team on our sales coaching solutions page. When enablement teams can see both individual and systemic gaps, they can put coaching time where it will matter most.
FAQ
What is a sales coaching gap?
A sales coaching gap is a difference between the selling behaviors an organization expects and the behaviors its sellers consistently demonstrate. For example, a team may expect every rep to explore business impact in discovery, while assessments show many reps stop at surface-level questions. Gaps can belong to one rep, one team or the whole organization.
How do you prioritize sales coaching gaps?
Prioritize sales coaching gaps by prevalence, business relevance and coachability. A gap that affects many sellers, matters to the sales motion and improves with practice and feedback goes first. A gap limited to one group gets targeted help, and a rare or low-impact gap can wait until the next assessment cycle.
Can AI identify sales coaching gaps?
AI can help identify sales coaching gaps by scoring practice conversations against a shared rubric and showing patterns across sellers, skills and cohorts. AI roleplays give every seller the same scenario, which makes results easier to compare. Calibrate the scores with human reviewers and check them against real-call evidence before acting on them.
How often should sales teams assess coaching gaps?
How often depends on the sales cycle, onboarding cadence and business priorities. Many teams assess after each onboarding cohort and after major training initiatives, messaging changes or methodology updates. Reassess the same skills with the same criteria after each round of coaching so you can see whether sellers actually improved.
-
How to Run an AI Fluency Assessment Across Your Org
An AI fluency assessment shows where your org stands with AI tools, role by role, before you spend a dollar on training. Most companies start with a guess. They assume engineering is fluent, sales is behind, and everyone else sits somewhere in the middle. A structured assessment replaces that guess with data, so enablement can build the right training tracks, leadership can set realistic targets, and the whole org can see progress over time. This guide walks through how to run one that measures real skill.
If your team has already picked a tool, our guide on how to train your team to use Claude at work covers the rollout side. This post covers measurement.
What an AI Fluency Assessment Measures
A usage survey tells you how often people open a tool. To learn whether they use it well, you need to assess AI skills across employees on four components drawn from Anthropic’s AI Fluency course and its 4D framework.
- Delegation: the person picks the right tasks for AI and keeps the right ones for themselves.
- Description: the person gives the tool enough context and structure to get a usable result.
- Discernment: the person evaluates the output critically and catches errors.
- Diligence: the person applies the company’s data policy correctly and owns what they send.
Self-report surveys miss all four, because people rate their own skill generously. Performance tasks ask people to do the work, and then you score the result.
Step 1: Define AI Fluency by Role
AI fluency for a support rep looks different from AI fluency for a legal ops lead. The assessment has to reflect the real work of each function, or the results won’t tell you much.
For each role group, list the five to eight tasks where an AI assistant should be part of the workflow. Here is a starting point:
- Sales: pre-call research, follow-up drafting, and objection prep.
- Support: ticket summaries, response drafting, and root-cause hypotheses.
- Marketing: first-draft copy, competitive summaries, and FAQ generation.
- Legal ops: contract summaries, policy explainers, and process docs.
Those task lists become the assessment content. If you need a sense of what proficient looks like in each function, our AI fluency benchmarks by role give you targets to score against. The same principle applies when you are building AI roleplays around your actual sales methodology, because the practice has to look like the job.
Step 2: Design Three Task Types for Each Role
A complete AI competency assessment uses three task types. Each one maps to the four components, so together they give you a full picture.
This table shows which task type measures each of the four AI fluency components and how that task is scored.
Component What it shows Task type How it is scored Delegation Picks the right tasks for AI and keeps the right ones Production task Rubric for accuracy, completeness, tone, and format Description Gives the tool enough context and structure for a usable result Production task Rubric for accuracy, completeness, tone, and format Discernment Evaluates output critically and catches errors Verification task Planted errors caught and fixed Diligence Applies the data policy correctly and owns what they send Judgment task Accuracy sorting inputs as allowed, allowed with changes, or not allowed Production Tasks
The employee completes a real deliverable with the AI tool, using a realistic input the company provides. For example, a support rep gets a long ticket thread and drafts a resolution message. Score the output against a rubric for accuracy, completeness, tone, and format. Production tasks measure Description and Delegation.
Verification Tasks
The employee receives a plausible AI draft with planted errors and has to find and fix them. The errors might be a wrong product detail, a misstated policy, or an unsupported claim. Score on errors caught. Verification tasks measure Discernment, which is the component teams most often skip.
Judgment Tasks
The employee sorts a set of realistic inputs and requests under the company’s data policy. Each item is allowed, allowed with changes, or not allowed. Score on accuracy. Judgment tasks measure Diligence. If your policy owners want a shared reference point for governance language, the NIST AI Risk Management Framework is a common place to start.
Keep the whole assessment under 45 minutes. Long assessments get rushed or abandoned, and either one corrupts the data.
Step 3: Run the Assessment on a Consistent Platform
Spreadsheets and email work for a small pilot. At scale you need a platform that gives every employee in a role the same tasks, scores everyone against the same rubric, and reports by role, team, and component.
Yoodli built its Tool Training templates with an assessment component for this purpose. Admins pick a template for the tool in question, such as Claude for Workplace or Putting Copilot to Work. Then they customize the tasks and evaluation criteria for their org and assign it by role. The expanded Yoodli template library removes the design work that stalls most assessments before they start.
Each template pairs an AI Tutor session that teaches the skill with a mastery assessment that checks what people can actually do. Results roll up by component in Yoodli Measure. So enablement can see that a function is strong on production and weak on verification, then build for that.
Step 4: Treat the First Run as a Baseline
The first AI fluency assessment run sets your baseline. Hold off on leaderboards and consequences for now. The goal is a clear picture of where each function stands, and employees who fear the result tend to game it.
Say so from the start. Tell employees the company is measuring where it is today so it can build the right training, and that nobody’s rating depends on this round. Managers should get their team’s results as a coaching input. Performance conversations can come later, once the training has had time to work.
Step 5: Read the AI Skills Gap Analysis by Component and Role
The most useful view is a grid. Put roles down the side and the four components across the top, with a score in each cell. Patterns show up quickly.
- High production and low discernment means a team uses the tool a lot and trusts it too much. Verification training is the fix.
- Low production and high judgment means a team is careful but gets little value from the tool. Better examples and context-setting practice will help.
- Low judgment across the board points to how the data policy was communicated. Start with clearer policy training, then retest.
This AI skills gap analysis gives enablement a build list. It also gives leadership one view of workforce AI readiness across the company.
Step 6: Build Training Tracks From the Gaps, Then Reassess
The AI fluency assessment tells enablement what to build. A role that needs discernment gets a track heavy on verification scenarios. A role that needs description gets context-setting practice on real tasks, and a role that needs diligence gets policy scenarios with clear right answers.
Build the tracks, run them, and reassess the same roles in 60 to 90 days. If you already run onboarding and certification programs, the same mechanics apply here. Certify on demonstrated skill, then track it.
The change between baseline and reassessment is the number leadership wants to see, because it shows the training built a real capability. In Kirkpatrick model terms, that is Level 2 and Level 3 evidence, which most AI adoption programs never collect. The Kirkpatrick Partners overview explains each level if your team needs a refresher.
Use an AI Ride-Along Coach to Assess AI Fluency on Real Work
A tutor-led certification confirms that each person can complete a set task. Ride-along sessions come after it, once people know the basics and are applying the tool to their own work. In Yoodli, a ride-along is a type of AI roleplay. The employee shares their screen while working in Claude, ChatGPT, Microsoft Copilot, or Gemini, and the coach sees what they type and what the tool returns. The coach stays quiet until the employee asks a question. Then it gives one short suggestion based on your company’s guidance, and the employee keeps control of the work.
Take a support rep whose baseline showed low discernment. She brings a real ticket thread into Claude and asks for a draft reply to the customer. The draft mentions a refund window, so she asks the coach, “How do I check this before I send it?” The coach answers from the company’s guidance: confirm every policy detail against the current policy page before anything goes out. She finds the window is wrong, fixes the line, and finishes with a reply she can send.
Each ride-along is scored against two to four goals, which become its rubric. Good goals include prompt clarity, iteration, output review, and policy fit, which cover description, discernment, and diligence. Manager dashboards show playback, transcripts, and score trends for every session, and Yoodli Measure shows progress across programs and against your rubrics. That gives you assessment data from real work between formal reassessments. The transcripts also show which parts of your guidance need more detail.
Setup happens in the Roleplay Builder, starting from the Tutor template. An admin writes one plain-language description, and the builder drafts the cast, Context, and goals. Keep the coach’s guidance short. Then assign the ride-along in a Program after the tutor sessions so completion is tracked. The help center guide to building a ride-along roleplay covers the Context format.
Format When to use it AI Tutor lesson First exposure to a tool or feature, or a workflow everyone does the same way AI Tutor certification Proving that each person can complete a specific task AI ride-along coach After the tutor sessions, once people apply the tool to their own varied work Table: Run AI Tutor sessions first and ride-along sessions after, as laid out in Yoodli’s guide to the AI ride-along coach.
What This Looks Like in Practice
Enterprise teams already use practice-based assessment to measure readiness at scale. One example is how a global IT organization pressure-tested AI readiness with Advantexe and Yoodli. Another is Loopio, which cut demo certification attempts by 67% and certified 60 reps with AI Tutor.
The same pattern carries over to AI tools. People learn by doing the task, getting feedback, and doing it again. An assessment built that way gives you results you can trust. Our guide to AI platform certification covers how to turn those results into a formal credential.
Common Mistakes to Avoid
A few mistakes show up again and again when companies measure AI adoption.
- Teams measure usage alone. Weekly active users is an adoption metric, and it says nothing about quality or risk.
- Teams use one test for everyone. Role-specific tasks are what make the results useful to each function.
- Teams skip discernment. Many orgs test whether people can produce with AI and never test whether they can catch its errors.
- Teams run it once. A baseline only becomes useful when you have a second data point to compare it to.
FAQ
What is an AI fluency assessment?
An AI fluency assessment is a role-based test of how well employees use AI tools in their real work. It scores four components: delegation, description, discernment, and diligence. Employees complete realistic tasks from their own jobs, so the results show actual skill. Enablement teams use the results to plan training by role.
How do you assess AI skills across employees?
You assess AI skills by giving each role group realistic production, verification, and judgment tasks, then scoring them against a shared rubric. Run every employee in a role through the same tasks on one platform. Report results by role and component so you can see which skills each team needs to build next.
How long should an AI fluency test take?
An AI fluency test should take under 45 minutes for each employee. That is enough time for a few production, verification, and judgment tasks without causing fatigue. Longer tests tend to get rushed or abandoned, which skews the data. Short, focused tasks tied to real work give cleaner results.
How often should you reassess AI fluency?
Reassess AI fluency 60 to 90 days after the first training tracks launch. That window gives people time to practice and apply new habits. After the first reassessment, many teams move to a quarterly or twice-yearly rhythm. Each new round shows whether skills are holding and where new gaps have appeared.
Start With a Baseline
A structured assessment turns a vague goal into a plan with clear targets by role. Define the tasks, test all four components, read the results as a grid, and build training from the gaps. Yoodli gives L&D and enablement teams the templates, AI Tutor sessions, mastery assessments, and ride-along sessions on real work to run that whole cycle in one place.
-
Claude Training for Healthcare Teams: A Practical Guide
Health systems, payers, and life sciences companies are buying enterprise AI seats quickly, and Claude training for healthcare teams has to keep pace with that rollout. A hospital operations group gets Claude licenses. A few early users start drafting policy summaries and staff communications in the first week. Many others open the tool once, feel unsure about what is safe to type, and go back to their usual workflow. A good program settles the safety question first, builds the skill on real tasks next, and then proves both with an assessment.
This guide lays out a rollout plan for clinical operations, revenue cycle, patient access, and administrative teams. It draws on how Yoodli helps healthcare organizations train staff on new tools with AI tutoring, AI roleplays, and mastery assessments. If you are new to the topic, our general guide on how to train your team to use Claude at work covers the basics that apply to every industry.
Start Claude Training for Healthcare With the PHI Boundary
Every healthcare training program on an AI tool begins with one question: what can go into the tool, and what has to stay out? Protected health information is governed by HIPAA. Whether a specific Claude deployment can handle PHI depends on your contract, your configuration, and whether a business associate agreement is in place. Your compliance team owns that answer. Training makes sure every user knows it well enough to apply it without stopping to ask.
The first module should be a short, scenario-based check. Show the learner a realistic task, such as “summarize this patient’s visit notes for the referral.” Ask them to decide whether the input is allowed under your policy. If it is not, they explain how to de-identify it or route the task another way. The learner makes a real decision and gets scored on it, which builds the habit much faster than a policy slide.
This step gives users confidence. People who know exactly where the line sits use the tool more often, because they stop second-guessing every prompt. It also gives your compliance team a record that every user was tested on the policy before getting access.
Map Claude Use Cases by Healthcare Role
Healthcare has many distinct workforces under one roof. The tasks a revenue cycle analyst hands to Claude look very different from a nurse manager’s tasks. Build Claude training for healthcare around three or four role tracks instead of one generic course.
The table below compares the drafting and summarizing tasks each healthcare role track typically hands to Claude.
Role track Drafting tasks Summarizing and explaining tasks Administrative and operations staff Internal communications and first drafts of SOPs Long policy documents and meeting notes turned into action lists Revenue cycle and patient access Appeal letter templates Denial codes in plain language and payer policy updates Clinical leadership and educators Training materials and talking points for hard team conversations Clinical guidelines turned into staff-facing summaries Life sciences commercial teams Field talking points within approved claims and internal briefings Published literature Administrative and operations staff
These teams use Claude to draft internal communications, summarize long policy documents, turn meeting notes into action lists, and prepare first drafts of standard operating procedures.
Revenue cycle and patient access
These teams use it to explain denial codes in plain language, draft appeal letter templates, and summarize payer policy updates for the rest of the department.
Clinical leadership and educators
Leaders and educators turn clinical guidelines into staff-facing summaries, draft training materials, and prepare talking points for hard team conversations.
Life sciences commercial teams
Commercial teams summarize published literature, prepare field talking points that stay within approved claims, and draft internal briefings for brand and market access leads.
Each track gets its own examples, its own practice scenarios, and its own definition of a correct, policy-compliant output. Yoodli’s expanded template library for tool training includes a Claude for Workplace template. Admins customize the personas, evaluation criteria, and session content for each role. Because the structure is prebuilt, a training lead can spend most of the setup time on department examples. The same setup supports other clinical and frontline programs, as our overview of AI training for healthcare organizations shows.
Train Healthcare Staff on Real Tasks
Walking through the interface rarely makes anyone better at the job. Healthcare employees get value from Claude through a handful of repeatable habits. They give it the right context and ask for a specific format. They check the output against the source, and they know when a task is a poor fit for the tool.
Start with Yoodli’s AI Tutor, which walks learners through the tool end to end using realistic tasks from their own role. A patient access supervisor might practice drafting a script that explains a prior authorization delay. Then they get feedback on whether the draft is accurate and empathetic, and whether anything went into the prompt that should have stayed out. Practice on real work is what moves people from trying the tool once to using it every week. Our AI tool training page shows how the same approach works for other enterprise tools.
Conversation skills matter too. Staff who use Claude to prepare a hard message still have to deliver it to a patient or a colleague. Our post on AI roleplays for healthcare teams covers how organizations rehearse those moments. Ochsner Health’s frontline leadership program is one example of that practice running inside a health system.
Build a Mastery Assessment for Healthcare AI Adoption
Completion tracking tells you who sat through the training. To know who can use Claude safely and well, you need a mastery assessment. Each Yoodli Tool Training template pairs the tutor session with an assessment that evaluates what learners can actually do.
A strong healthcare assessment has three parts:
- In the judgment section, the learner sorts five realistic tasks into allowed, allowed with de-identification, or not allowed.
- In the production section, the learner completes two real tasks in Claude and submits the output.
- In the verification section, the learner reviews an AI-generated draft with a subtle error and has to catch it.
Learners who pass all three sections earn certification in Claude training for healthcare. Learners who miss the judgment section retrain before they get access, even if their drafts were strong. This gating approach mirrors how onboarding and certification programs work for other high-stakes skills, and it scales. RingCentral cut call-center certification time by 90% with Yoodli, which shows how much review time automated assessment can return to a training team.
Use an AI Ride-Along Coach for Claude on Real Healthcare Work
Daily healthcare work varies more than any lesson plan, so the step after certification is an AI ride-along coach. The coach stays with an employee while they work in Claude on their own tasks. The employee shares their screen, and the coach sees what they type and what Claude returns. When the employee asks a question, the coach gives one short, specific suggestion from your organization’s guidance on approved use cases, prompt standards, and data handling. It stays quiet the rest of the time and never takes over the work.
Here is a typical session. A revenue cycle analyst wants an appeal letter template for a denial code her team sees every week. She opens the session in Yoodli, shares her screen, and asks Claude for a first draft. Then she asks the coach, “Can I paste in the denial letter we received?” The coach answers from the data rules the admin added to its setup. It suggests using the payer’s published policy text and the denial code, and leaving the letter out because it contains member details. She revises the prompt and finishes with a reusable template for that denial.
Run the AI Tutor and the ride-along coach in order: tutor sessions first, then ride-along sessions once staff know the basics and apply Claude to their own work.
Format When to use it AI Tutor (first) First exposure to Claude or a new feature, a workflow everyone does the same way, and a certification that proves each person can do a task AI ride-along coach (after) Staff who know the basics and are applying Claude to their own, varied work, with one short answer from your guidance when they ask Table: Use the AI Tutor first and the ride-along coach after, as our guide to what an AI ride-along coach is explains.
In Yoodli, a ride-along is a type of AI roleplay. You build it in the Roleplay Builder from the Tutor template by writing one plain-language description, and the builder drafts the cast, Context, and goals. Keep the coach’s guidance short and set two to four goals, such as prompt clarity, output review, and policy fit. Assign it in a Program after the tutor sessions so completion is tracked. The help center guide to building a ride-along roleplay covers each step. Transcripts show what staff ask the coach, so your compliance team can see where the guidance needs more detail.
Run Claude Training as an Ongoing Competency Program
Healthcare organizations already run competency programs. Annual compliance training, skills validation, and credentialing all follow a familiar rhythm. Claude training for healthcare teams fits the same model and deserves the same level of rigor.
Assign the training by role, with a deadline. Report completion and pass rates to department leads. Recertify when your policy changes or when the tool changes in a meaningful way, which happens often with AI. Everboarding and continuous enablement follow the same logic, because ongoing practice matters more than the first rollout.
Financial services teams work with a similar mix of regulation and fast-moving tools. Our guide to Claude training for financial services covers how those teams handle client data and certification, and much of it transfers directly to payers and life sciences companies. For organizations with large frontline and guest-facing workforces, our guide to Claude training for retail and hospitality teams shows how to fit the same program between shifts.
Measure Healthcare AI Adoption After Certification
Adoption is the leading indicator for Claude training for healthcare teams. Look at active use by role a month after certification. If clinical educators use Claude weekly and revenue cycle staff use it rarely, the revenue cycle track probably needs better examples. Interview a few users and rebuild the scenarios around the tasks they handle most.
The lagging indicators are operational. Track the time to produce an appeal letter draft, the turnaround on policy summaries, and staff satisfaction with internal communications. Yoodli Measure shows assessment scores and practice activity by team, so you can connect training results to those outcomes. For a broader evaluation framework, the Kirkpatrick Model applies cleanly to tool training.
Compliance and quality leaders often want a shared vocabulary for AI risk as the program grows. The NIST AI Risk Management Framework gives them one, and it pairs well with the judgment scenarios in your assessment.
FAQ
Can healthcare staff put patient information into Claude?
It depends on your organization’s deployment, contract, and policy. Whether Claude can handle protected health information comes down to how your organization configured it and whether a business associate agreement is in place. Your compliance team should make that call. Training should then teach every user to apply it, including how to de-identify information when the policy requires it.
How long does it take to set up Claude training for a healthcare team?
Setup time depends on how many role tracks you build and how much practice each one includes. With a prebuilt template, a training lead can set up several tracks quickly and customize them with real examples from each department. Learners then complete a tutor session and a mastery assessment, and the program adds short refreshers whenever the policy or the tool changes.
What should a healthcare AI training assessment include?
A healthcare AI training assessment should include judgment, production, and verification. The judgment section tests whether learners know which data is allowed. The production section has them complete real tasks in Claude. The verification section asks them to catch an error in an AI-generated draft. Gate access on the judgment section so every certified user handles patient data correctly.
Which healthcare roles benefit most from Claude training?
Administrative, operations, revenue cycle, patient access, and clinical education teams usually see value first, because their work involves frequent drafting and summarizing. Life sciences commercial teams also benefit when training keeps outputs within approved claims. Start with the role that has the most repetitive writing tasks, then expand once the first track shows strong adoption.
Bring Claude Training to Your Healthcare Teams
Set the PHI boundary first and test for it. Build role-specific tracks, train on real tasks with the AI Tutor, and certify with an assessment that covers judgment, production, and verification. Then support daily work with ride-along sessions, and recertify as the policy and the tool change. Yoodli, the AI Platform for Human Skills, gives healthcare training leaders the template, the AI Tutor, the assessment, and the ride-along coach in one place. Staff build safe, confident habits with Claude in their everyday work.
-
AI Roleplay Security: What Enterprise Buyers Should Ask Before Rollout
AI roleplay security comes down to five questions. Ask what the platform stores, who can see it, whether anyone trains AI on it, how long it lives, and which audits prove it. Ask for a document or a setting behind every answer, and bring your security team in before the pilot starts.
This guide is for enablement and L&D leaders who own the program, and for the IT and security reviewers they bring in. It lists the questions that separate a quick review from a stalled one, and what a good answer looks like for each. We build Yoodli, the AI Platform for Human Skills, so we close with how we answer the same questions. For the rest of the buying criteria, such as realism, feedback quality and reporting, read our guide on how to evaluate AI roleplay platforms.
Why AI Roleplay Security Gets More Scrutiny Than Other Training Tools
AI roleplays hold more sensitive data than a typical course. A rep’s practice session can include a voice recording, a transcript, a scored rubric, and the pricing or competitive talk track the scenario was built from.
Look at what goes into a single AI roleplay program. Enablement teams upload playbooks, discovery guides and objection handling notes so the AI buyer sounds real. Learners record their voice, and sometimes video, for every attempt. Managers then review scores tied to named employees. That mix of company IP, recordings and performance data is why a security team treats an AI roleplay platform differently from a slide library.
The wider threat picture raises the stakes. IBM and the Ponemon Institute put the global average cost of a data breach at a record high in the Cost of a Data Breach Report 2026. They also track attacks that use AI as their own category.
Figure Finding 4.99M Global average cost of a data breach, in USD, up 12% on last year 56% Increase in AI-driven attacks 6M Global average breach cost, in USD, of an AI model inversion attack 1.93M Cost savings, in USD, from extensive use of AI and automation in security Breach costs reached a record high in 2026, and AI-driven attacks rose 56%. Source: IBM, Cost of a Data Breach Report 2026
Your reviewers will ask hard questions about any AI vendor. A program owner who brings the answers early gets through review faster.
Which Certifications Should an AI Roleplay Platform Have?
An enterprise AI roleplay platform should have a current SOC 2 Type 2 report at minimum, plus clear answers on GDPR and, for teams in Europe, the EU AI Act. ISO certifications and AI risk frameworks add assurance, but the SOC 2 scope matters more than the logo count.
SOC 2 is the American Institute of Certified Public Accountants (AICPA) standard for reporting on a service organization’s controls. A SOC 2 report covers controls relevant to security, availability, processing integrity, confidentiality, or privacy, and the vendor picks which of those criteria are in scope. A Type 1 report checks control design at one point in time. A Type 2 report tests whether the controls worked over an audit period, which is why security teams ask for it by name.
Most established vendors in this category now publish a SOC 2 Type 2 report. Treat it as the entry ticket, and spend your review time on what the report covers and what it leaves out.
AI adds frameworks that a classic security review never touched. ISO/IEC 42001 sets requirements for an AI management system. The NIST AI Risk Management Framework, released on January 26, 2023, is intended for voluntary use. A vendor doesn’t need every badge, but it should be able to say which frameworks shape how it governs its AI.
Term What it tells a buyer SOC 2 Type 2 Controls worked over an audit period, tested by a CPA firm SOC 3 A public, shorter summary of a SOC 2 examination ISO/IEC 27001 The vendor runs a certified information security management system ISO/IEC 42001 The vendor runs a certified AI management system NIST AI RMF A voluntary US framework the vendor may use to manage AI risk EU AI Act EU law that sets duties for AI systems by risk level Each certification answers a different question, so ask what each one covers before you count them.
The AI Roleplay Security Checklist: Questions to Ask
Send these questions to every vendor on your shortlist before the demo stage. A good answer names a document, a setting or a contract clause. A vague answer is a sign the review will run long.
Question to ask What a good answer looks like Does your SOC 2 Type 2 report cover the product we would buy? A current report, shared under NDA, with the roleplay product in scope Do you or your AI model providers train on our data? A written no for all organization data, backed by the contract What data retention terms apply to your AI model providers? Enterprise agreements with minimal retention and no training Who can see a learner’s recordings? Private by default, shared by the learner or by admin roles you control How long do recordings and transcripts live? An admin-set retention window, plus deletion when the contract ends Do you support SSO and SCIM? SAML or OpenID Connect sign-in you can enforce, with automated provisioning Is data encrypted in transit and at rest? Yes for both, with the controls listed in a Trust Center Does any feature infer emotions, and how are scores used? A clear answer mapped to the EU AI Act, reviewed with your legal team A good security answer points to evidence, such as a report, a setting or a contract clause.
What Data Does an AI Roleplay Platform Collect and Store?
Ask for the full list: account details, recordings, transcripts, uploaded content and scores. Then ask where it is hosted and which subprocessors touch it. A published subprocessor list with update notifications saves your privacy team a round of email.
Your own team controls part of this answer. Build scenarios from fictional account names and sanitized deal details. The AI buyer still sounds real, and your most sensitive data never enters the platform. Score data needs the same care, because readiness reports in tools like Yoodli Measure tie results to named people.
Does the vendor train AI models on your recordings?
This is the question security reviewers ask most often about AI vendors, so get the answer in writing. The vendor should state that your organization’s data is never used to train AI models, by the vendor or by its model providers. Ask which plan tiers that promise covers, because some vendors exclude data from training only on paid business plans.
How are retention and deletion handled?
Look for a retention window your admin sets, after which recordings are deleted automatically. Ask what happens when the contract ends, how long deletion takes, and which records the vendor keeps for billing or support.
What identity and access controls are in place?
Enterprise rollouts depend on single sign-on (SSO) through your identity provider and automated user provisioning through SCIM. Ask how roles and permissions work, who can see whose recordings, and whether you can enforce SSO for every user.
What Does the EU AI Act Mean for AI Roleplay Security?
The EU AI Act matters if you have employees in the EU. Article 5 bans AI systems that infer emotions of a natural person in the areas of workplace and education institutions, except for medical or safety reasons. Annex III lists AI used to monitor and evaluate the performance and behaviour of workers as high-risk.
So ask two questions. Does any feature infer a learner’s emotions? And will practice scores feed decisions about promotion, pay or termination? Keep practice scores for coaching and certification, and have your legal team review how the scores are used. This article is not legal advice.
How to Run the Security Review Without Stalling the Rollout
Start the security review the same week you start vendor demos. Most delays come from a review that begins after the business has already picked a vendor.
- Collect the documents first. Ask each finalist for its SOC 2 Type 2 report, penetration test summary and any completed standard questionnaires, such as a CAIQ or HECVAT. Many vendors share these through a Trust Center after an NDA.
- Send the checklist above. Your security team can then focus on gaps rather than starting from a blank questionnaire.
- Pilot with low-risk content. Run the AI roleplay pilot on fictional scenarios while the review finishes, so the pilot and the review run in parallel.
- Plan delivery and identity together. Map SSO groups to programs and decide how practice reaches learners. Our guide to SCORM vs. LTI covers the LMS side.
- Agree on retention before launch. Set the retention window and the score-sharing rules before the first cohort records anything.
Our enterprise AI roleplay rollout guide covers the phases that follow, from rubric calibration to program expansion.
How Yoodli Answers These Questions
Yoodli publishes its security answers in the Yoodli Trust Center, so reviewers can check them without waiting on a sales call. Here is how we answer the checklist.
- Audits and compliance. Yoodli is SOC 2 Type 2 certified, GDPR compliant and EU AI Act compliant. The Trust Center offers our SOC 2 Type II report, SOC 3 report, penetration test, CAIQ, HECVAT and Responsible AI Policy on request.
- AI training. All organization data, including users’ data, is never used by Yoodli or our subprocessors for AI training or prompt improvement. Our pricing page shows that business plans exclude data from AI training by default.
- Privacy and encryption. All recordings are private by default until the user shares them. Data is encrypted in transit and at rest.
- Retention and deletion. An admin can set an organization retention policy, and recordings are deleted after the set number of days. When an agreement ends, Yoodli deletes the organization’s product data within 60 days.
- Identity. Yoodli supports SSO and SCIM, with role-based access controls.
Here is how that looks in setup. An admin sets the retention window in organization settings before launch. The enablement team then writes each scenario’s instructions to steer the AI persona away from topics the company wants kept out of practice.
The same setup runs at enterprise scale. Google Cloud used Yoodli to certify 15,000+ employees on a new GTM pitch, and the training scored a 92% CSAT.
FAQ
Is SOC 2 Type 2 enough for AI roleplay security?
SOC 2 Type 2 is necessary but rarely enough on its own. It shows that a vendor’s controls worked over an audit period. You still need to confirm the roleplay product is in scope, that your data never trains AI models, how long recordings are kept, and how SSO and SCIM are handled.
Do AI roleplay platforms train AI models on your recordings?
Some do and some don’t, so ask for the answer in writing. A good vendor states that organization data is never used to train AI models, by the vendor or its model providers. Check which plans the promise covers, because some platforms exclude data from training only on paid business plans.
Which AI roleplay platforms are SOC 2 Type 2 compliant?
Most established AI roleplay platforms, including Yoodli, publish a SOC 2 Type 2 report. Because the certification is common, compare vendors on what the report covers. Ask for the current report under NDA, confirm the roleplay product is in scope, and check the dates of the audit period.
Does the EU AI Act apply to AI roleplay training?
It can, if you train employees in the EU. The Act bans AI that infers emotions in the workplace, except for medical or safety reasons. It also treats AI that evaluates worker performance for employment decisions as high-risk. Ask vendors how their features map to the Act, and involve your legal team.
Bring Security In Early
The fastest AI roleplay rollouts treat security as part of the evaluation from day one. Send the checklist, collect the documents, and pilot on fictional scenarios while the review runs. To walk through Yoodli’s answers with your security team, talk to our team.
-
Difficult Conversations for New Managers: How to Prepare Them
The best way to prepare first-time managers for hard conversations is to give them a framework, then have them practice each conversation out loud before it happens for real. Workshops and reading lists cover what to say, but a new manager under pressure falls back on whatever they have rehearsed.
This guide is for L&D and HR leaders who run manager development programs. It covers the difficult conversations for new managers that come up most and how to prepare for each one. It also covers where AI practice fits next to executive coaching and how to measure readiness. We build Yoodli, the AI Platform for Human Skills, so we use it as the worked example and say so when we do.
Why New Managers Struggle With Hard Conversations
Most new managers were promoted for doing their old job well, and that job rarely included telling a peer their work is slipping. The skills that carry a hard conversation are emotional and verbal. They only build through use, and a new manager has had almost no chance to use them.
The discomfort is common. A Harris Poll survey for the communications firm Interact asked 616 U.S. managers which parts of the job made them uneasy. Over a third (37%) said they are uncomfortable giving direct feedback about an employee’s performance that the employee might respond badly to. Smaller shares were uneasy with tasks that sound simple, like recognizing good work.
Communication task Managers uncomfortable Giving direct performance feedback the employee may react badly to 37% Demonstrating vulnerability 20% Recognizing employee achievements 20% Delivering the company line in a genuine way 20% Giving clear directions 19% Speaking face to face rather than by email 16% Critical performance feedback is the conversation managers are most uncomfortable having. Source: Harris Poll for Interact, 2016
Training before the promotion helps, but it is far from universal. The Cegos international barometer surveyed 4,271 first-time managers in 10 countries and found that 56% were trained before assuming the role. The same survey found 47% lack time for the human aspect of their role, and 68% of HR respondents said internal candidates lack soft skills.
HR leaders see the strain too. In Gartner’s 2025 HR priorities research, three-quarters of 805 HR leaders reported their managers are overwhelmed by the expansion of their responsibilities. Gartner also found that 75% of organizations have made significant updates to their leadership development programs, yet they are not seeing results.
Which Difficult Conversations for New Managers Are Hardest?
The hardest conversations for a first-time manager are the ones where the relationship and the result pull in different directions. Four come up again and again: performance feedback, ongoing underperformance, conflict between team members, and saying no.
Each one asks something different of the manager. A program that treats them as one generic “hard conversation” skill misses that. The table below is a starting map for scenario design.
Conversation What the new manager has to do Performance feedback Name the behavior and its impact without making it personal Underperformance Set a clear expectation, a timeline and a follow-up Conflict between peers Hear both sides and move them to a shared next step Saying no Explain the priority and keep the person’s trust Delivering unwelcome news State the decision plainly and stay for the reaction Managing a former peer Reset the relationship without pulling rank Each hard conversation calls for a different skill, so each deserves its own practice scenario.
Conflict deserves extra attention. UK research from the CIPD found that people managers are just as likely to exacerbate a conflict situation as they are to resolve it. That is a training problem L&D can address directly, because conflict conversations follow patterns a manager can rehearse.
How to Prepare a New Manager for a Specific Conversation
Preparation for a single conversation has three parts: the facts, the opening, and the likely reactions. A manager who has rehearsed all three walks in calmer and leaves with a clearer outcome.
Start with facts. The manager writes down two or three specific, dated examples and the impact each one had on the team. This keeps the conversation on behavior instead of character. A model like SBI (Situation, Behavior, Impact) gives new managers a simple structure for this step.
Then script the first two sentences. Many new managers lose control of a hard conversation early, either by apologizing for having it or by talking around the point. A clear opening sentence fixes most of that.
Finally, plan for the reactions. Employees get defensive, go quiet, cry, push back with their own grievances, or agree too quickly. Each reaction needs a different response from the manager, and this is the part that reading cannot teach. It takes practice against a realistic counterpart.
Why Practice Matters More Than Content in Difficult Conversations Training
Practice is what lets a new manager use the right words when the other person is upset. Much difficult conversations training stops at content, because live practice is hard to run at scale.
Traditional practice depends on a facilitator or a peer playing the employee. That works in a workshop, but it rarely happens again before the real conversation. Peers also tend to go easy on each other, so the manager never meets the defensive employee they are dreading.
Our earlier post on new manager training lays out a 12-week program built around practice. This post goes one level down, into the conversations themselves. The crucial conversations use case shows how teams set those conversations up in Yoodli.
VML is a good example of practice at scale. Its L&D team built a leadership series called Hi to help 26,000 employees across 50+ markets get better at difficult conversations. The team teaches a framework first and follows it with guided, customized practice in AI roleplays. The Hi Series improved leader confidence by 20 percent, and 81 percent of participants reported feeling prepared for their toughest conversations.
What AI Tools Help New Managers Practice Difficult Conversations?
AI roleplay platforms let a new manager rehearse a specific conversation with an AI counterpart, then get scored feedback against the organization’s own rubric. The manager can repeat it as often as needed, privately, before the real meeting.
When you compare tools for a manager development program, look for these capabilities:
- Scenarios built from your real situations, language and leadership model, rather than a generic library.
- Counterparts that react differently, so a manager practices against a defensive employee and a quiet one.
- Scoring against your framework, such as SBI or GROW, with feedback the manager can act on the same day.
- Progress data by learner, cohort and program, so L&D sees who is ready.
- Delivery inside the tools managers already use, such as your LMS.
Here is how that looks in Yoodli. An L&D team builds AI roleplays for each conversation type and grades them against rubrics it defines. It sets up AI personas with their own appearance, voice and context, so a manager can face a skeptical direct report one day and a blunt one the next.
Ochsner Health ran this kind of program for entry- and mid-level leaders. Its team built twelve custom goals with subject matter experts and used three AI personas: enthusiastic, blunt and skeptical. Leaders practiced performance, career development and trust-restoration conversations. The Ochsner case study reports that the goals with the largest improvement were among the lowest starting scores.
AI Coaching vs. Executive Coaching for First-Time Managers
AI practice and executive coaching do different jobs, and first-time managers do best with both. A coach helps a manager decide what to say and why. AI roleplays give the manager the repetitions to say it well, between coaching sessions and before the real conversation.
Executive coaching is also scarce. Many organizations reserve it for senior leaders and high-potentials, so many first-time managers never get a coach at all. AI practice extends some of that support to every new manager, and it gives coaches better material to work with. Our post on pairing executive coaching with practice covers how the two fit in one program.
What AI roleplay practice Executive coaching Best for Rehearsing a specific conversation Judgment, mindset and career decisions How often As often as the manager wants Scheduled sessions Who gets it Every new manager in the program Usually senior leaders and high-potentials Feedback Scored against your rubric right away A coach’s read on the whole situation Strongest when Used between coaching sessions Informed by the manager’s practice data AI practice gives every new manager repetitions, and executive coaching adds the judgment a coach brings.
If you are building a shortlist, our guide to AI leadership coaching platforms covers what to test in a demo.
How L&D Measures Manager Readiness for Hard Conversations
Readiness is measured by skill scores on realistic practice, tracked over time, alongside manager confidence and team outcomes. Completion rates and workshop satisfaction scores only show who attended.
A practical readiness model for difficult conversations for new managers uses three layers:
- Practice scores. Each manager’s scores on each rubric goal, per conversation type, from the first attempt to the latest.
- Self-reported confidence. A short pulse before and after the program asking how prepared managers feel for their toughest conversations.
- On-the-job signals. Direct-report feedback, engagement scores and regretted attrition on the manager’s team over the following quarters.
Practice data is the layer most programs lack. Yoodli Measure shows progress across programs and against rubrics, so L&D can see which skills need more investment and which managers need more support. Ochsner’s team says this data now shows where to focus development resources across a large system.
BDO Gisborne shows what this looks like in a leadership program. After adding AI practice to its manager and leadership training, BDO reported a 25% increase in the participants’ management capabilities, including coaching and giving feedback. It also found that 100% of participants reported improved confidence.
How to Roll Out Difficult Conversations Training for New Managers
Start small, with one cohort and the two or three conversations your new managers avoid most. A focused pilot gives you clean before-and-after data and real scenarios to improve before a wider rollout.
- Pick the conversations. Ask HR business partners which conversations go badly most often for first-time managers.
- Bring subject matter experts in early to write the scenarios and the rubric in your own language.
- Teach the framework first, then assign practice for each conversation type.
- Have managers practice before their real conversations, and again after, so the practice follows the work.
- Review the practice data with coaches or program leads at the midpoint, and adjust scenarios that are too easy.
- Report skill improvement and confidence to the business at the end of the cohort.
Then expand. Teams that start with one manager cohort often add more conversation types, more cohorts and more regions once the first results are in. Our L&D solutions page covers how Yoodli supports that kind of program.
FAQ
What are the most difficult conversations for new managers?
The most difficult conversations for new managers are performance feedback, ongoing underperformance, conflict between team members and saying no. Managing a former peer and delivering unwelcome news are close behind. Each calls for a different skill, so strong programs give each conversation its own scenario and rubric instead of teaching one generic approach to hard conversations.
What AI tools help new managers practice difficult conversations?
AI roleplay platforms help new managers practice difficult conversations by letting them rehearse with an AI counterpart and get scored feedback right away. Look for scenarios built from your real situations, personas that react differently, scoring against your framework, progress data for L&D, and delivery inside your LMS. Yoodli supports each of these for manager development programs.
Is AI coaching a substitute for executive coaching for first-time managers?
AI coaching works alongside executive coaching for first-time managers by giving them practice between sessions. A coach helps with judgment, mindset and career decisions. AI roleplays let the manager rehearse a specific conversation as often as needed. Many first-time managers never get an executive coach, so AI practice extends support to every new manager.
How do L&D teams measure readiness for difficult conversations?
L&D teams measure readiness with three layers of data: skill scores on realistic practice, self-reported confidence and on-the-job signals. Practice scores show improvement on each rubric goal over time. Confidence pulses show how prepared managers feel. Direct-report feedback, engagement and regretted attrition show whether the change holds up with the team.
Give New Managers Practice Before the Real Conversation
New managers will have hard conversations in their first months whether they are ready or not. L&D can make sure the real conversation is a manager’s fifth attempt instead of the first. Start with one cohort and the conversation your managers avoid most, and let coaches build on what the practice data shows.
-
DevLearn 2026 Recap: AI, Practice, and Real Behavior Change
In this DevLearn 2026 recap, we share what L&D leaders told us in Las Vegas, what our customer VML said on the Expo Stage, and what it all means for 2027 learning plans. The question we heard most wasn’t “should we use AI in learning?” It was “how do we prove it changed what people actually do?”
For three days, November 4 to 6, the Yoodli team was at Booth 662 in the MGM Grand Expo Hall. We talked with L&D leaders, enablement teams, and learning designers about where AI fits in their programs, and where it doesn’t.
DevLearn 2026 recap: VML on the Expo Stage
On Thursday, Yoodli CRO Josh Vitello and Kelly Stuart-Johnson, Global Head of Learning at VML, took the Expo Stage for “AI Won’t Make You a Better Leader.” The session looked at what AI really changes in leadership development, and what it doesn’t.
VML is a 26,000-person global agency. Its leadership program, Hi, gives managers a place to rehearse difficult conversations with AI roleplays before they have them for real. The idea behind it is simple: most managers don’t lack knowledge. They lack reps.
The session covered four ideas L&D teams can use now:
- Confidence is not competence. Training often builds the feeling of readiness without the skill. Practice closes that gap.
- AI practice makes reps cheap. A manager can run a tough feedback conversation five times in a lunch break, with no peer awkwardness and no stakes.
- AI can help teach empathy. Realistic practice partners surface how a message lands, not only what it says.
- Use the time AI saves on people skills. Research from UC Berkeley Haas found that AI tools tend to fill the time they free up with more tasks. Teams that want better conversations have to protect that time on purpose.
“Our managers knew what a good feedback conversation looked like. What they didn’t have was a safe place to practice one before it counted.” — Kelly Stuart-Johnson, Global Head of Learning, VML
Want the full story? Read how VML won four Brandon Hall awards with Hi, built in partnership with Yoodli.
What L&D leaders told us at Booth 662
Over coffee and ice cream, the same three themes kept coming up. They run through this whole DevLearn 2026 recap.
1. Completions aren’t the metric anymore. Leaders want to show behavior change: a manager who handles a hard conversation better, a rep who ramps faster, a new hire who’s ready sooner. Course completions don’t answer the question their executives are asking.
2. AI fluency is now a training program, not a memo. Many teams are rolling out AI tools company-wide and finding that access doesn’t equal adoption. People need practice using AI in their real work, with feedback.
3. Practice has to live in the flow of work. A one-time workshop fades fast. Teams want realistic AI roleplays before the moment that matters, and coaching right inside the tools people already use.
One L&D director at a national healthcare system put it plainly: “My CFO doesn’t care how many people finished the module. She wants to know if our nurse managers are having better conversations.”
A sales enablement lead from a SaaS company added: “We gave everyone an AI license in March. By June, half of them had never opened it.”
A CLO’s view from the floor
DevLearn capped a busy first month for Mike Mather, Yoodli’s new Chief Learning Advisor and former Chief Learning Officer at KPMG. In three weeks he went from CLO Exchange in New York to HR Tech to DevLearn.
“A year ago, CLOs asked me whether AI belonged in learning. This month, every one of them asked how to prove it’s working. That’s the right question, and it starts with measuring what people do, not what they complete.” — Mike Mather, Chief Learning Advisor, Yoodli
After hours with the L&D Coalition
Thursday ended at Topgolf Las Vegas with the L&D Coalition happy hour. Thank you to everyone who came out to swap notes, swing a few clubs, and keep the conversation going.
Key takeaways from our DevLearn 2026 recap
If this DevLearn 2026 recap has one message, it’s that L&D teams are done measuring activity. They want proof that people got better at the work.
Yoodli is the AI Platform for Human Skills. Enterprise teams use it to onboard employees, train people on AI tools, run certifications, and put a coach inside their workflow. The goal is the same one we heard all week: measurable behavior change, not course completions.
Missed us in Vegas? Talk to our team and we’ll show you how teams like VML put practice to work. Next up, catch us at Sales Enablement Collective in Toronto, November 12 to 13.
Happy speaking!
DevLearn 2026 recap FAQ
When and where was DevLearn 2026?
DevLearn 2026 Conference & Expo ran November 4 to 6, 2026, at the MGM Grand in Las Vegas.
What were the biggest themes at DevLearn 2026?
Measuring behavior change instead of completions, building AI fluency across the workforce, and moving practice into the flow of work.
How does VML use AI to train managers?
VML’s leadership program, Hi, lets managers at the 26,000-person agency rehearse difficult conversations with AI before having them for real, using Yoodli. Read the VML story.
How can L&D teams measure behavior change, not just completions?
By assessing how people perform in realistic practice scenarios over time, then tracking whether those skills show up in real work. Yoodli scores practice against the skills a team defines, so leaders see progress, not just participation.
What is Yoodli?
Yoodli is the AI Platform for Human Skills. Enterprise teams use it to onboard employees, train people on AI tools, run certifications, and put a coach inside their workflow.
-
Multilingual Sales Training for Global Teams and Partners
Multilingual sales training lets sellers and partners learn, practice, and get certified in the language they use with customers. Most global programs are built at headquarters in English and shipped out to the regions. A rep in Tokyo, São Paulo, or Munich learns the pitch in English, then translates it live in front of a buyer.
This guide is for enablement leaders running programs across regions. It covers why practice language matters, where programs fall short, and what to check in an AI roleplay platform. It also covers how to keep one scoring standard as you roll out. Last, it shows how Yoodli supports multilingual AI roleplay for direct sales teams and partners.
Why multilingual sales training matters
Multilingual sales training matters because buyers prefer their own language, and reps sell better when they don’t have to translate on the call. A rep who rehearses in English and sells in German is doing two jobs at once: translating and selling.
Buyers notice the language of the conversation. In a CSA Research survey of 8,709 consumers in 29 countries, 76% of online shoppers said they prefer to buy products with information in their native language. Business buyers feel the same way. In CSA Research’s look at B2B buyers, nearly two-thirds (66%) of business users said they’d pay up to 30% more for a localized product.
Speaking is also where second-language gaps show most. The 2025 EF English Proficiency Index drew on 2.2 million test takers across 123 countries and regions. Speaking was the weakest English skill in over half of the countries measured. Many reps can read an English slide deck with no trouble and still struggle to handle an objection out loud in English.
Cultural context matters as much as vocabulary. Directness, small talk, and the way buyers raise concerns all differ across markets. A rep who practices with a buyer persona built for their market learns the rhythm of the conversation along with the words.
Where multilingual sales training programs fall short
Most global sales training is built in one language and then translated. Translation usually stops at the content, so practice stays in English or doesn’t happen at all.
- Practice left in English. Reps read localized modules, then practice in English or skip practice entirely.
- Live roleplay limited by trainers. Regional teams need trainers who speak the language, so practice depends on who is free that week.
- Inconsistent standards. Each region scores practice its own way, so “ready” means different things in different places.
- Slow updates. When the pitch changes, every translated version has to be rebuilt, and regions fall behind headquarters.
AI roleplays address the first two directly. The AI persona plays the buyer in the rep’s language, so reps practice whenever they need to without waiting for a local trainer. Yoodli has supported roleplay and feedback in languages beyond English since its multi-language roleplay launch in 2024.
Area Translated content only Multilingual practice Practice Often in English or skipped Roleplays in the rep’s selling language Feedback In English, if any In the language the rep practiced in Scoring Varies by region One global rubric Scenarios Headquarters examples Local buyers and objections Launch timing Regions wait for translation Regions launch on the same schedule Translation alone stops at the content, while multilingual practice carries the rep’s language into practice, feedback, and scoring.
Multilingual training for partner teams
Partners need multilingual sales training most, because they carry a large share of IT sales and often sell in local languages. Omdia’s 2026 analysis of IT spending trends puts the partner share of the total addressable IT market well above half in every region.
Region Partner share of IT market, 2026 Latin America 78.1% EMEA 72.2% Asia Pacific 68.6% North America 61.1% Partners account for more than 60% of IT spending in every region, and the most in Latin America and EMEA. Source: Omdia, 2026
If most of your Latin American revenue runs through partners, your partner certification has to work in Spanish and Portuguese. Partners often get the least localized training of any group, even though they reach many international customers. A multilingual partner certification holds partner reps to the same standard as your direct team, in the language they sell in.
Yoodli’s partner enablement programs can run on a partner portal through web embed or in a partner LMS through SCORM or LTI. SAP is a useful example of a vendor planning for this kind of reach. SAP has pledged to equip 12 million people with AI-ready skills by 2030. Its February 2026 update describes widening partner-led training for broader geographic coverage and languages. Yoodli supports SAP’s certification experience through SAP Learning.
Which AI roleplay platforms support practice in multiple languages for global teams?
Many AI roleplay platforms now list a language count, so the count alone won’t settle the decision. The better question is which parts of the experience change when a learner switches languages. Ask every vendor to walk through each layer below in the demo, using a language you don’t speak at headquarters.
Language layer What to ask in the demo Spoken practice Does the persona speak, listen, and give feedback in that language? Regional variants Can learners pick the regional version, like Spanish for Spain or Latin America? Platform interface Do buttons and menus follow the learner’s language? Admin content Who translates roleplay titles, descriptions, and goals? Scoring Does the same rubric score every language? Persona instructions Can admins write scenario instructions in any language? A multilingual roleplay platform has to handle language in several layers, and spoken practice is only the first.
A few more checks belong on your list:
- Coaching in the rep’s language. Feedback that arrives in English after a session in Japanese sends the rep back to translating.
- Demand in your growth markets. Yoodli has seen Korean, Japanese, and Thai roleplays grow fast, because live coaching in those markets is expensive and often not localized.
- Enterprise readiness. Global programs touch employee data in many countries, so check security certifications, GDPR, and admin provisioning early.
- One platform for every team. Direct sales, partners, and support should share one platform, so one rollout doesn’t need three vendor reviews.
How to keep one scoring standard across languages
Keep one standard by grading every language against one rubric and one set of talking points, then comparing results by region. When every region uses the same scenarios and rubric, a certified rep in Singapore and a certified rep in Chicago met the same bar.
Start with the rubric your best reps already meet. Write the criteria as behaviors (“confirms the customer’s priority before pitching”) rather than phrases. A phrase translated word for word can read stiff or wrong in another language, while a behavior holds across cultures. Then let the AI score the behavior in whatever language the rep uses.
Google Cloud shows how much one rubric can carry. Its team built Google Cloud scenarios, custom rubrics, and required talking points, then certified 15,000+ employees on a new GTM pitch at 92% training CSAT. The same structure works when reps speak different languages. The scenario and rubric stay fixed, and the conversation happens in the rep’s language.
Then watch the data. Yoodli Measure shows progress across programs and against rubrics, so you can break results out by team. If one region scores well below the rest on one criterion, review a handful of its recordings before you change anything. The fix may be coaching, or the scenario may need a local detail.
Track participation by language too. If reps in one language practice far less than others, check whether the scenario feels natural in that language. Small fixes, like a familiar buyer name or a local competitor, often help.
Localize the scenarios along with the language
Good localized training adapts the situation as well as the words. Buying processes, common objections, and competitive context differ by market. A procurement conversation in Germany may focus on different concerns than one in Brazil.
Yoodli Personas let you build local buyers with the role, temperament, and objections that fit each market. Ask regional sales leaders for the top objections they hear and build them into the persona. Reps then hear the pushback they’ll actually get on their next call.
Bring regional leaders in early. Ask each region to review the core scenarios, suggest local objections, and nominate a few reps for the pilot. Keep the rubric and core scenarios global, and let regions add local scenarios on top. A region selling heavily to the public sector, for example, might add a procurement scenario other regions don’t need.
How to roll out multilingual sales training region by region
Roll out one region at a time, starting where the need is clearest. Reuse the same scenarios everywhere you can, and fix localization problems in one market before they reach ten.
- Pick a pilot region. Choose one market with a clear goal, like a launch or a certification deadline.
- Localize the scenario. Adjust names, competitors, and objections for that market.
- Translate learner-facing text. Translate titles, descriptions, and goals before launch.
- Set the practice language. Have reps pick their language before the first session.
- Review early recordings. Have a regional manager check that scores match their judgment.
- Expand by region. Reuse the scenario and rubric in the next market.
Launch timing is where many programs struggle. Headquarters finishes the English version, and regions wait weeks for localized training. Yoodli Create builds a full learning journey from a prompt and your docs, calls, and methodology. That shortens the time to build each version, so regions don’t fall as far behind.
Consistent messaging across regions is often the real goal. A global cybersecurity company used Yoodli AI roleplays to keep product messaging aligned across regions as features shipped. Its sellers practiced on their own schedule across time zones.
Plan delivery around the tools reps already use. Many teams embed Yoodli in their LMS so regional reps find practice next to the rest of their training. Share your rollout plan with security and legal early, too. Yoodli’s SOC 2 Type 2 certification and GDPR compliance usually come up in global reviews.
Coaching after the training
Coaching after the program keeps new skills from fading. Yoodli Coach reviews real calls from Gong and Kaia, scores them against your rubric, and turns the pattern into the rep’s next practice. That works for regional teams as long as their calls are recorded in one of those tools.
Because the rubric is global, managers in every region coach from the same data. A manager who moves from one region to another already knows how to read the results.
How Yoodli handles multilingual AI roleplay
Yoodli is the AI Platform for Human Skills, and it supports AI roleplays in 25+ languages. The rep picks a practice language before the session, and that choice drives the transcript, the persona’s replies, the AI voice, and the post-session feedback.
Yoodli’s language and localization guide lists the supported languages, including regional versions of English, French, Portuguese, and Spanish. For example, reps can choose Spanish for Spain or Spanish for Latin America, and Portuguese for Brazil or Portugal. Admins can write persona instructions in any language. The interface follows the rep’s browser language.
Admin-written content, such as roleplay titles and goal descriptions, stays in the language it was written in. Admins who need it in several languages can ask their Yoodli CSM to turn on admin translation, which shows each rep the version for their language. Build that step into your rollout so reps don’t see an English title above a Japanese roleplay.
FAQ
What is multilingual sales training?
Multilingual sales training lets sellers and partners learn and practice in the languages they use with customers. Strong programs localize practice and certification along with the content. They also keep one scoring standard, so readiness means the same thing in every region and leaders can compare teams fairly.
What is the difference between multilingual AI roleplay and translated training?
Translated training swaps the text on the screen into another language. In a multilingual AI roleplay, the rep speaks with an AI persona in their own language and gets feedback in that language. That tests whether they can handle a live buyer or partner conversation, which a translated course or quiz can’t measure.
Can AI score roleplays fairly across different languages?
Yes, if every language is scored against the same rubric written as behaviors. Criteria like “confirms the customer’s priority” translate cleanly, while exact phrases don’t. Review a sample of recordings in each new language during the pilot. Have a regional manager confirm the scores match their own judgment before you scale.
How many languages should an AI roleplay platform support?
Support every language your customers and partners buy in, plus the regional variants that matter. Coverage of your markets matters more than the total count. Check that feedback, scoring, and regional versions like Latin American Spanish work in each one. Yoodli supports 25+ languages, including regional versions of English, French, Portuguese, and Spanish.
Start with one market
To prove a multilingual program fast, pick one region with a real deadline. Build one scenario and score it with the rubric you already trust. Once the pilot region’s scores match what its managers see on live calls, roll the same scenario to the next market.
-
8 Things to Know About Integrating Yoodli Into Your LMS
If you stopped by our booth at DevLearn, there’s a good chance you asked us some version of “Does Yoodli work with our LMS?”
The short answer is yes, and the longer answer is below. We rounded up the eight things L&D teams ask most about a Yoodli LMS integration. Together they show how Yoodli roleplays fit into the learning stack you already have (no rip and replace required).
1. It Speaks Your LMS’s Language
Yoodli supports SCORM 1.2 plus LTI 1.1 and 1.3, the common standards learning platforms run on. If those acronyms blur together, our SCORM vs. LTI explainer breaks them down. Teams are already using Yoodli with Docebo, Moodle (MoodleCloud too) and Allego.
2. Learners Never Have to Leave the Course
Download a SCORM package for any Yoodli scenario and upload it to your LMS. Learners then open the AI roleplay right inside the course they’re already taking. They don’t have to juggle extra tabs or wonder where to log in.
3. Scores and Completions Report Back on Their Own
When a learner finishes a roleplay, Yoodli sends completion and scoring data straight back to your LMS. Your reporting stays in one place, and nobody has to export a spreadsheet at 5pm on a Friday.
4. Pick Your Flavor: SCORM or LTI
SCORM is the quick, simple route. LTI takes a bit more setup but brings extra muscle:
- Learners get sorted into groups for their practice content.
- Scores still land if someone closes the activity early.
- Human grading is supported, so managers or coaches can weigh in.
Not every LMS supports LTI, so check yours first. Either way, we’ll help you choose.
What you get SCORM 1.2 LTI 1.1 and 1.3 Setup Download a package and upload it to your LMS Configure Yoodli and your LMS together LMS support One SCORM integration works across platforms Not every LMS supports LTI New learners Auto sign-up adds them to one group you pick Group provisioning adds new and existing users to a group Score reporting Sent when the learner submits Holds up if a learner closes the activity early Human grading Score doesn’t update after a grader changes it Supported SCORM is the faster setup, and LTI adds group provisioning and human grading, according to Yoodli’s LMS integrations help article.
5. One Roleplay Can Power a Whole Curriculum
With LTI, you map LMS courses or activities to Yoodli roleplays, and several activities can point to the same roleplay. Build a great discovery call scenario once. Then drop it into onboarding, a refresher course and your manager track. Intellum customers see the same pattern in their own platform, as our post on AI roleplays inside Intellum shows.
6. No More Roster Wrangling
With LTI, Yoodli can add learners to your org and the right group automatically the first time they launch a roleplay from your LMS. That works for brand new users and for people who already have an account, once your email domain is verified with Yoodli. SCORM has its own version, called auto sign-up, which helps when you’re onboarding new hires in big waves.
7. Cameras and Mics Are Covered
Roleplays mean talking, and some LMSs lock down camera and mic access inside embedded content. The fix is usually a quick setting, like opening the activity in a new window in Docebo. Our support team has the playbook for each platform.
8. The Yoodli LMS Integration Comes With Enterprise (Plus a Lighter Option)
LMS integrations come with Yoodli’s Enterprise plan. Not ready to connect an LMS? Yoodli Programs gives you built-in, LMS-style ways to assign and track practice without any integration at all. Teams weighing other options can also read our roundup of LMS alternatives for sales training.
Ready to Plug In?
Your LMS is where learning lives. Yoodli is where people practice the conversations that matter. Put them together and practice becomes part of every course.
Want to see it working in your LMS? Book a demo and we’ll walk you through it. Already a customer? The full setup guide is in our LMS integrations help article.