You Ran the Survey. Then You Emailed the Results.

Internal AI surveys are everywhere right now. Ask about usage, ask about blockers, tally it up, put it in slides. That part usually gets done carefully. What comes next is almost always the same: circulate the results company-wide, and stop.

Our position, stated plainly: measuring is not the goal. A survey only becomes real work when the people in it walk out knowing which role their team is missing. Stop short of that and all anyone receives is an opinion about some numbers.

This is written for people who run these surveys and for people who get made to fill them in. If you have ever taken one, somewhere in the middle of this you will find yourself wondering what actually happened to your results. That question is the whole subject.

What happens when you stop at visibility

A survey has stages. Individuals see their own result. The group sees its distribution. And then people notice which role is missing. Circulate-and-stop almost always halts at stage two.

What fills the gap there is either blame or indifference. Hand someone the single fact that adoption is low and their available moves are to suspect their colleagues of not caring, or to decide this isn't their problem. Neither one produces a next action.

Carry it to stage three and the conversation changes shape. The moment a team sees that nobody among them brings in new ways of working, it stops being anyone's failing and turns into an unfilled seat. Seats can be assigned without accusing anyone.

The 60 minutes we actually run

The assessment is sometimes run inside a training session. Here is that agenda. The minutes below are the design allocation, not a measured record, so read them that way.

• Ten minutes of theory first. Establish why different people suit different roles. Skip this and the assessment lands out of nowhere
• Ten minutes taking it, in the room. Not sent home. Take-home versions lose respondents, and the missing respondents distort the distribution
• A fifteen-minute break. It is a break and it is also the tallying window. Cut it and results become "we'll email those out later", which is a different exercise entirely
• Twenty minutes looking at results and talking. This is the actual session. In order: your own result, the group's distribution, the missing roles
• Five minutes deciding who takes what away.

Three things make it work. One is teaching the theory before measuring anything. Another is spending the break on tallying, so that taking it and seeing the results close on the same day. Put a week between those two and the people who answered have moved on.

The third matters most: three things get said out loud before anyone answers a question. No type is better than another. Nobody has to cover all of them. Roles are not fixed and do change. Leave those unsaid and people answer while wondering whether this feeds their performance review. Answers given under that suspicion produce a discussion where nobody talks about themselves.

The moment that lands is the empty cell

In the results session, what moves a room is never the averages or the rankings. It is an empty cell in the grid of roles against types. We call those white spaces.

One caution belongs with it. Small denominators produce empty cells by chance. A team of seven with zero of some type is not evidence that the type is absent. So a white space is not a conclusion, it is the next thing to check. Present it as settled and somebody in the room will start talking about hiring before lunch.

The four questions that come up every time

Facilitate this and you get almost the same four questions on every run. Without prepared answers, the room stiffens right here. These are ours.

All four need answers ready in advance. The first one especially: hedge on it for two seconds and you can feel the room close.

Move one thing this week

If you already ran a survey and the results have been sitting in an inbox since, there is exactly one thing to do. Take those results and lay them out again by role, showing who is where. No new questionnaire needed. Any gap on that sheet is the first seat to fill.

If the survey is still ahead of you, block thirty minutes after the tallying to look at the results together. Whether a room reads them before they get circulated changes what the same survey is worth.

Our assessment is 24 questions and takes about four minutes, free, no signup. But taking it is only half of this. Put the fifteen types next to it and count how many seats on your team are actually occupied. That is the point where a survey turns into work.