# How to Measure AI Capability Across Your Workforce

Why logins and prompt counts mislead, and how to measure whether your people are genuinely getting better with AI.

**To measure AI capability, track how well people use AI in their real work against a defined scale, not how often they log in.** Usage metrics tell you activity; they say nothing about whether anyone is producing better work. The organisations getting real value measure movement in ability over time, not volume of prompts.

## Why the usual metrics mislead

Most AI reporting counts the wrong things:

- **Licences activated** tells you who has access, not who is capable.
- **Prompts per user** rewards volume, not quality. Someone can send fifty shallow prompts a day and never change how they work.
- **Course completions** tells you who sat through training, not who applied it.

All three can look healthy while capability, and value, stays flat.

## What to measure instead

Measure capability the way you would measure any skill: against defined levels, observed in real work.

1. **Establish a baseline** for each person, based on what they can actually do with AI.
2. **Define levels** so progress has a shape. On the five-level Atheni Scale, people move from Curious, to Capable, to Fluent, to Pathfinder, to Trailblazer.
3. **Look at real outputs**, not self-reported confidence. Capability shows up in the quality and ambition of the work, not in a survey.
4. **Track movement over time**, by person, team and role, so you can see who is progressing and who needs support.

## The one number that matters

The single most useful measure is **how many people are moving up a level over a defined period**. That is the real signal that capability is building. A team where most people climb a level in a quarter is transforming; a team with high login numbers and no movement is not, whatever the usage dashboard says.

## How Atheni does it

Atheni maps every person onto the Atheni Scale, embeds role-specific guidance in their real work, and reports movement up the scale over a 90-day window, giving leaders, for the first time, a view of the *quality* of AI use across the organisation rather than just the quantity. Across sectors, this approach has consistently reached **over 90% adoption within 90 days**.

## The bottom line

If your AI dashboard can tell you who has logged in but not who is getting better, you are measuring the wrong thing. Measure capability against a scale, in real work, over time. To see where your workforce sits today, [book a demo](/demo/).

Atheni Strategy Ltd, London. Contact: hello@atheni.ai.