Reporting AI search visibility to your board or members

Boards and members have started asking a question that did not exist a few years ago: what do the AI assistants say about us when someone asks about our issue? It is a fair question, because a growing share of journalists, staffers and members now put that question to ChatGPT, Perplexity or Google’s AI answers before they reach a person. The honest answer is that you cannot look it up — there is no public ranking to check — but you can observe it, and report it plainly. The right report measures four things, fits on one page a non-specialist can read, shows a baseline and a trend rather than a flattering snapshot, and ties back to the outcome the mandate is judged on. Here is how to build it.

Why boards and members now ask about AI answers

For a coalition or an association, the website was never the point — being the trusted source on your issue was. For years that meant search engines and the press. Increasingly it means the assistant a person asks first. When a member’s general counsel, a committee staffer or a reporter wants a quick read on your issue, a meaningful number of them now type the question into an AI assistant and take the answer at face value.

That changes what a board needs to know. “How does our site rank?” is no longer the whole question. The board’s real question is whether, when the issue is raised with an assistant, the organisation is present, correctly described and central to the answer — or absent, misrepresented and crowded out by other voices. Reporting that is not vanity; it is accountability for the core mandate.

The four things worth measuring

Most of the confusion around AI visibility comes from trying to reduce it to a single number. It does not reduce cleanly, and a board is better served by four readings than by one invented score.

• Presence — are we mentioned? When your defining questions are put to the assistants, does your organisation come up at all? This is the floor. If you are not present, nothing else matters yet.
• Accuracy — is it right? When you are mentioned, is what the assistant says about you correct? Wrong figures, an outdated position or a garbled mandate are worse than silence, because they travel with apparent authority.
• Source coverage — who is cited? When an assistant supports an answer with citations, whose pages does it lean on — yours, a peer body’s, an opponent’s, a reference source it trusts? This tells you whether you are part of the evidence the machine reads, or merely hoping to be.
• Share of answer — us versus others. When the issue is answered, how much of the answer is your organisation and your framing versus other named voices? This is the AI-era share of voice: present and central, or present but peripheral.

Together these four say something a single figure never could: not just are we there, but are we right, are we cited, and do we own the room.

A one-page board format a non-specialist can read

The instinct is to hand the board a dense dashboard. Resist it. The board does not need the method; it needs the picture. A page that works has four parts.

• The four readings, each as a plain line. Presence, accuracy, source coverage and share of answer — one sentence each, in words, not jargon. “On our ten defining questions, we are now mentioned in eight, up from five at baseline.”
• The baseline beside the latest reading. Every line shows where you started and where you are now, so the trend is visible at a glance and nothing rests on a single snapshot.
• One representative example, quoted honestly. A short, real extract of what an assistant actually says when asked your central question — not the best answer you could find, the typical one. This makes the abstract concrete for a non-specialist.
• One line on what it means for the mandate. Translate the readings into the outcome the board cares about: are we the source on our issue, yes or no, and is that improving?

If a reader who has never heard the term “generative engine optimisation” can read that page in a minute and know whether things are getting better, it is doing its job.

Baseline and trend, not vanity numbers

The single most important discipline is to fix a baseline and report movement from it. Before any work begins, record exactly how the assistants answer your defining questions today — including the questions where you are absent or wrong. That uncomfortable baseline is the most valuable number in the whole exercise, because everything afterwards is judged against it.

It matters that the question set stays fixed. If you quietly swap in easier questions, the report will improve while reality does not, and you will have built precisely the vanity dashboard a board is right to distrust. Hold the same questions, ask them of the same assistants, and let the trend be real.

Because there is no public ranking to quote, this is measured by observation, not lookup. You take a fixed set of the questions your audience actually asks, put them to the main assistants on a regular cadence, and record what comes back. Answers vary between runs and shift as the assistants update, so any one reading is noisy — which is exactly why the report should lead with the direction of travel over time, never a single flattering result. We set out the underlying method in how to measure AI search visibility.

What good progress looks like across a quarter

Over a quarter, “good” is rarely a clean sweep — it is steady, defensible movement on the four readings. Presence rises: you appear on questions where you were previously absent. Accuracy firms up: the assistants describe your mandate and positions correctly, because the correct version is now well published and easy to read. Source coverage shifts toward you: your own pages, and authoritative third-party references on your issue, start showing up among the citations. Share of answer grows: on your central questions, your framing is present and central rather than a footnote to someone else’s.

Just as telling is what good refuses to claim. It does not pretend every question improved at once, quote one strong answer as though it were the rule, or dress observation up as a precise score. A report that only ever goes up is not a triumph; it is a warning that someone is choosing the questions. Honest progress includes the places you are still weak and says plainly what is being done about them.

Tying visibility back to the mandate

A board does not fund AI visibility for its own sake. It funds the mandate — winning the argument, defending the position, being the body people turn to on the issue. So the report’s last move is always to connect the four readings to that. If your coalition exists to be the authoritative voice on a specific rule, the question is simple: when someone asks an assistant about that rule, are you the voice it reaches for, and is that truer this quarter than last? When visibility is framed that way, it stops being a technical curiosity and becomes what it should be — evidence on whether the organisation is fulfilling the job it was set up to do.

This is the companion to how to measure AI search visibility and why discoverability is the barrier to trade association growth. For the wider shifts behind it, see strategic communications trends for 2026.