Anonymous or identifiable: what your organisation sees, and where the limit is

People answer candidly when they know who will see what they say.

Updated 6 October 2026. We first published this post on 21 April 2026, when every pulse was anonymous. Each pulse is now anonymous or identifiable, and the post describes how that works today.

Every survey has the same failure mode. You ask a question people have strong feelings about, you promise their answers are confidential, and somewhere between the survey form and the leadership deck the names quietly travel too. People notice, and the next time you ask, the candid answers are gone.

We've watched this play out in employee engagement surveys for decades, and we did not want to repeat it for AI. Our first response was to show every answer without a name, on every pulse.

That turned out to be too blunt. Some pulses work better with names on them: a pulse about where people want help is of little use if you can't offer that help to the person who asked for it, and in a small organisation people are often happy to put their name to what they say. So the choice now sits with each pulse.

Each pulse says which it is before it starts#Link to this section

Every pulse is either anonymous or identifiable, and says which before its first question. Your organisation sets the default for new pulses, and a new organisation starts with anonymous.

When a pulse is anonymous, your name isn't shown with your answers to anyone, admins included. Owners, Admins, Senior Leaders and Group Leaders see averages, totals, trends and comments word for word (a Group Leader, for the groups they lead), and every figure needs answers from at least five people before it appears. Owners and Admins can also read single answers and the day each was given, with no name shown. Senior Leaders and Group Leaders see the figures and comments, never the answers one by one.

Identifiable pulses say so before you start, and show your name with your answers. Owners and Admins see your name beside your answers, and so does whoever leads your group. You can close an identifiable pulse and answer nothing.

What an Admin sees on an anonymous pulse#Link to this section

Here is a comment from an anonymous pulse, as an Admin sees it in Reports:

"Copilot's autocomplete keeps suggesting deprecated Python 3 APIs."
Copilot · 2/5

The comment, the tool and the rating are all there. The name isn't, for the Admin or for anyone else in the organisation.

Hiding a name on the screen while it sits in the data behind the page would be a weak promise, because anyone with the browser's developer tools could read it. So on an anonymous pulse the name comes off on our servers, before the answer is shown anywhere. Each anonymous answer in the Reports list arrives in the browser looking like this:

{
  "id": "q8f2c1...",
  "user": { "id": null, "name": null, "email": null },
  "aiTool": { "name": "Copilot", ... },
  "answers": [ ... rating, time saved, comment ... ],
  "disclosureMode": "anonymous",
  "answeredOn": "2026-09-14",
  "createdAt": null
}

There is no name, email or ID for the person. The answer's own ID is replaced as well, because it records the moment the answer was given, and the time of day comes off for the same reason; only the date stays. The organisation's export leaves the name off anonymous answers too.

Where the limit is#Link to this section

The GAiGE knows which answers are yours. That is how your AI Adoption Portal shows you your own answers. What an anonymous pulse decides is who is shown your name with them: nobody, admins included.

A name that isn't shown still leaves two ways for an answer to point to you, and the extension, the answer page and your AI Adoption Portal all say so: "In a small group, or from something you write in a comment, someone may still be able to work out an answer is yours."

A small group. The five-answer minimum rules out the plainest case, a team of one or two. In a team of six, though, people know each other, and someone who puts a figure next to what they know about their colleagues may still make a good guess.

What you write. Comments are shown word for word. If you write "as the only person in finance using Jasper", the words give you away with no name attached. If you'd rather a comment couldn't be traced to you, keep it about the tool rather than your own situation.

Why we chose clarity over more rules#Link to this section

Before settling on this, we built a stricter version. It compared every figure a person could see with every other, so that no two could be subtracted to leave a handful of answers, and it held back whatever failed the test.

It did the job, at a price. Figures waited for each week or month to finish. Results for teams that overlapped were held back, and small teams lost the most. Every report became harder to read. And no set of counting rules can close off every way of working out who said what.

We would rather an organisation can see what its small teams are saying, so we took the stricter rules out. On an anonymous pulse, every figure still needs answers from at least five people, and instead of more machinery we tell people where the limit is. Our view is that people would rather know the limit than be told there isn't one.

In short#Link to this section

Each pulse says before it starts whether it is anonymous or identifiable. When a pulse is anonymous, your name isn't shown with your answers to anyone, admins included, and every figure needs answers from at least five people. In a small group, or from something you write in a comment, someone may still be able to work out an answer is yours, and we would rather you heard that from us.

For team members: Who sees my answers? and Your say shapes how AI is used at work. For leaders agreeing how your organisation uses AI: Who should decide which AI tools are used? and the other Leadership articles. The methodology behind the figures we show: The 2.5× rule.