Skills gap assessment through AI roleplayFind the skills gaps in how people talk. Not in how they rate themselves.
Employees complete AI roleplay scenarios built from your own situations: a price objection, a missed deadline, an angry customer. Each conversation is scored against your competency model, and the scores add up to a skills map for every team.
- Acknowledges the problem−0.2
- De-escalation−1.4
- Clear next steps−0.9
- Saying no politely−1.1
- Product knowledgeMet
Why roleplay
Surveys show what people think they can do. Roleplay shows what they do.
Most skills gap analyses rely on self-ratings and manager ratings. Both are useful, and both measure opinions. Communication skills are easier to see in a conversation: the same situation, the same criteria, for everyone.
How confident people feel
Quick to run. People rate themselves on the skills they understand, not on how they actually behave under pressure.
What the manager has seen
Depends on what the manager happened to observe, and on each manager’s own scale.
What people say in the situation
Everyone handles the same conversation. Each answer is scored on your criteria, with the transcript as evidence.
Five steps to a training plan
How to run a skills gap analysis with AI roleplay.
- Step 1
Define the skills each role needs
Start from your competency model, sales playbook or service standard. For each role, pick the five to eight skills that matter and the level you expect.
- Step 2
Choose the conversations that show them
A skill is visible in a situation: a price objection, a missed deadline, a customer who wants a refund. Pick one or two situations per skill.
- Step 3
Measure the current level
Everyone completes the same AI roleplays. Each conversation is scored against your criteria, so the evidence is the same for every person and every team.
- Step 4
Compare with the required level
Gap equals required level minus current level, for each skill, person and team. The biggest gaps across a team show where training will pay off.
- Step 5
Close the gaps, then measure again
Assign practice on the weakest skills, then run the same scenarios again. Same scenarios, same criteria: the change is comparable.
A skills gap analysis, worked through
A support team of 40. Five skills, one priority.
Every agent completed three roleplays with an AI customer. Levels are on a 1–5 scale; the current level is the team median.
| Skill | Required | Current | Gap | Action |
|---|---|---|---|---|
| Acknowledges the problem | 4 | 3.8 | 0.2 | No action |
| De-escalation | 4 | 2.6 | 1.4 | Priority: two roleplays a week with an angry customer, for four weeks |
| Saying no politely | 4 | 2.9 | 1.1 | Practice refusing a refund outside the policy, with an alternative |
| Clear next steps | 4 | 3.1 | 0.9 | Add a closing criterion to every scenario |
| Product knowledge | 3 | 3.4 | — | Above the required level |
Skill, required level, current level, gap, action. Download it as a spreadsheet file, or let the platform fill it in from roleplay scores.
Individual report
Where self-ratings and roleplay disagree. That is the gap.
Each person sees their own self-rating next to their roleplay score, with the moments in the transcript that explain it. Managers see the same for the team.
- Acknowledges the problem4.23.9
- De-escalation4.02.4
- Saying no politely3.82.7
- Clear next steps4.13.3
- Product knowledge3.63.7
AI and expert assessorsr = 0.80correlation between AI scores and expert scores on the same conversations
Want a skills map of one of your teams, built on your own competency model?
Book a demoWhen to run it
Before training, after training, before a promotion.
Before a training program
Know which skills to train and who needs it most, before you buy or build a program.
After a training program
Run the same scenarios again. The change in scores shows what the program did.
Before a promotion
See whether future managers can already give feedback and handle a difficult conversation.
Assessment centers
Scenarios built from your real cases, scored the same way for every participant.
FAQ
What HR and L&D teams ask us. Answered plainly.
Assessing candidates rather than employees: AI interviewer.
How do you perform a skills gap analysis?
Define the skills each role needs and the level you expect, choose situations where those skills show, measure the current level, and compare. The difference is the gap. Then train the biggest gaps first and measure again with the same method. The five steps in detail.
What is an example of a skill gap?
A support agent knows the refund policy but cannot calm an angry customer down enough to explain it. The knowledge is there; the conversation skill is not. A knowledge test would not show this gap. A roleplay with an angry customer would.
What does a skills gap analysis look like?
A table with one row per skill: required level, current level, gap and action. The support team analysis above shows five skills, one clear priority and what to do about each gap.
Is there a template for a skills gap analysis?
Yes. Download the template as a CSV file and open it in any spreadsheet. It has the same columns as the table on this page.
Can ChatGPT do a gap analysis?
It can help draft a competency model or summarize data you give it. For a gap analysis, everyone handles the same scenarios on the same criteria, so attempts are comparable and add up to a skills map for each team. A skills gap analysis needs evidence of the current level, and roleplay scored on your criteria produces that evidence.
How is this different from a skills test?
A test checks what people know. A roleplay checks how they use it in a conversation: what they say, in what order, and how they respond when the other person pushes back.
Does the AI judge emotions or appearance?
No. It scores what people say and how they handle the conversation. Emotions are modeled only for the AI character, as part of the scenario.
See the gaps in one of your teams. Before you plan the next program.
Book a short demo. We will load your competency model and run one of your scenarios with you on the call.