← Back to blog
Marketing · Colin Kleine ·

AI visibility: what it is and how to measure it

AI visibility is how your company shows up in AI-generated answers to the questions your buyers ask. Measure it with a fixed question set, saved responses and written rules for counting mentions and citations, then use what you find to decide what to fix.

Start with the buying decision

Take questions from sales calls and product evaluations. Use the words buyers actually use, even when they differ from the category name in your positioning deck. Note why each question is in the sample.

Keep discovery questions apart from questions that already name your company. A branded prompt tells you how the business gets described. It does not show that a buyer who has never heard of you would find you.

A measurement record another person can check

FieldWhat to save
QuestionExact wording and the buyer decision it represents
ContextSystem, date, account settings and market used for the observation
AnswerThe saved response and any source links shown
MentionWhether the answer names the company under your written definition
CitationWhether a source link points to a page your company controls
AccuracyThe claim checked, the supporting evidence and any correction needed
Follow-upA proposed action, an owner and a later check

Define the denominator before publishing a percentage

If a dashboard shows a visibility percentage, ask what it is a percentage of. It could be prompts, answers or repeated runs of the same prompt. Those are different units, so label the denominator and keep the records underneath.

Decide in advance how to count an answer that misses the question, a failed request or a duplicate prompt, and write the rule down before you compare periods. Change the exclusion rule and the headline moves while your actual presence stays the same.

Keep mentions, citations and accuracy separate

A mention can contain an error, and a citation can point to an unrelated page. Read what the answer says before you mark it a win.

For a SaaS company, check whether the answer describes the supported use case correctly. Save the exact sentence and the product evidence you checked it against. "Positive mention" in a spreadsheet gives the product owner nothing to correct.

Compare the same question set over time

Freeze a core sample for comparisons and keep new questions on a separate list. Show additions and removals in the report. If you add a market or a system, report it as a new view; do not fold it into the old series without saying so.

When a metric moves, read the answers behind it. A change justifies an investigation, but it does not prove a particular page edit caused it. Keep a log of what you published and when, so you can check the timing and other explanations.

Report what the observation means for the business

Say the report shows your company described correctly, but buyers still cannot find the integration explanation they need. The next job is to close that evidence gap. Reporting that "visibility improved" would hide the buying question you have not answered.

Track referral visits and enquiries separately, using your agreed analytics definitions. Ask sales whether prospects mention finding you through AI search, and keep those recollections apart from tracked events. Do not stretch either record into proof of everything that influenced a purchase.

End the review with a decision record

Use the AI visibility audit to check the evidence behind the report. The AEO strategy guide helps you decide which questions to work on first.

Keep a short note of the finding, the decision and who owns it. Anyone on the team should be able to say why you changed a page, left it alone or asked for more evidence. When the numbers move next quarter, that note tells you what you changed and why.

People also ask

Is a company mention the same as a citation? No. A mention names the business; a citation links to a source. Define and report them separately so readers know what was observed.

Can one prompt prove that visibility improved? No. It records one response under its recorded conditions. Use a fixed question set and read the answers behind it before drawing a wider conclusion.

Should branded and unbranded questions share one score? No, keep separate views so the difference stays visible. How an assistant describes a company you named is a different question from whether a buyer who has never heard of you finds it.

Hiring the marketing leader who will own the reporting? Brief Scalerr on the role or use the form below.

Hiring someone to own search?

Scalerr recruits marketing leaders for tech companies, from a first Head of Marketing to a CMO. Brief us on a marketing leadership hire, or use the form below.

Request a call

Brief us on a marketing leadership hire

By sending, you agree to be contacted by the Scalerr team about your enquiry.