What Your Team Says About AI, and What You Think They'd Say
Ten statements. You answer them as you believe the typical person here would answer. Then you send your team a link and they answer the same ten anonymously. When 5 people have replied, you see the gap.
Most leadership teams are wrong about this by a wide margin, and being wrong about it is expensive: budget, training and board answers all get decided on the belief, not the reality.
The check
How the anonymity actually works
Most anonymous surveys are anonymous by policy: the data exists, and someone has decided not to look at it. This one is anonymous by structure. When a team member submits, their answers are added into a running total and the individual response is never written down. There is no per-person row, so there is nothing to read, nothing to hand over, and nothing to leak. The team view has no name field, no email field and no free-text box, and the response count is the only thing that grows.
You can optionally give your own address when you create the link, so the results link reaches your inbox. That is worth doing: the link cannot be recovered if you lose it, because nothing else is tied to it. It is your address for your own session. Nobody answering is ever asked for one, and there is no per-person record for an address to attach to even if there were.
Results stay hidden until at least 5 people have answered, and that floor is enforced on the server rather than in the page, because a floor enforced in the page is a suggestion. In a nine-person company, an average computed from two responses identifies people.
One thing worth being straight about: the person who sends the link can see the total count going up. If your team is small enough that the timing of responses identifies people, this tool is the wrong instrument and you should have the conversation directly instead.
Why the gap is the number that matters
An AI readiness score tells you where you are. A gap tells you whether you can trust your own picture of where you are, which is the more useful thing, because every decision you make about AI rests on that picture. Industry surveys put executive confidence at around 78% against employee-supported reality closer to 23%. Your own number is more useful than either.
The headline number is a conversation opener. The value is in the per-statement breakdown: a company whose gap is concentrated on "the information AI needs is already available to it" has a context problem, and one whose gap sits on "I know what our rules are" has a policy problem. Those need completely different fixes, and only one of them is worth spending money on this quarter.
Want the whole picture rather than ten statements? Score your own AI readiness first, then book a call and I will tell you whether a full discovery across your exec team, your engineers and the rest of the business is worth doing, or whether you already know enough to act.
Join the Drift Digest
One short email every couple of weeks: practical fractional CTO and founder-AI notes, new AI tells as models change, and a quarterly nudge to re-measure your voice.
Something went wrong. Please try again, or email web@ctoondemand.co.uk.
Common questions
What is the AI reality gap?
It is the distance between what a leadership team believes about AI use in their company and what their staff actually report. Industry surveys put it at roughly 78% of executives believing they have a clear picture of AI usage against something closer to 23% supported by employee data. This tool measures your own version of that number rather than quoting the industry one at you.
Is the team survey really anonymous?
Yes, structurally rather than as a promise. Team answers are added into running totals the moment they arrive and no per-person record is ever written, so there is no row for anyone to read, subpoena or leak. The team view has no name field, no email field and no free-text box. Results stay hidden until at least 5 people have answered, because an average computed from two responses in a small team identifies people. The person who starts the check can optionally give their own address to have their results link sent to them, since that link cannot be recovered otherwise. That is their address for their own session, and it changes nothing about what is recorded for anyone answering.
How many people do I need for this to work?
At least 5 responses before any result is shown. Below that you see the response count and nothing else. Ten to thirty responses gives a genuinely useful read. Above eighty you are into the territory the paid AI Discovery Report is built for, where interviews and billing data matter more than a survey.
What do I do if the gap is large?
Resist buying a tool. A large gap usually means access, permission or context is in the way rather than capability: people cannot reach the AI, are not sure they are allowed to use it, or cannot give it the information it needs. Look at the individual statements with the widest gaps, because those name which of the three it is.
Found a gap you didn't expect?
That is the normal result, and it is worth half an hour of your time rather than a panic. Book a call and I will tell you which of the three usual causes it is, whether or not you ever hire me to fix it.