Six Questions to Ask Your AI Visibility Data

A visibility score is a thermometer. It tells you there's a fever. It does not tell you why.

The why lives underneath: every prompt, every answer an engine gave, every page it cited. Most teams never read that layer, because reading a few hundred AI answers is nobody's idea of a good afternoon.

Now your AI app can read them for you. With the Renown connector, Claude or ChatGPT can pull your own measurements and do the tedious part. The trick is asking questions that the headline number can't answer.

Here are six.

TL;DR

What changed, and on which engine? A headline move usually has one engine behind it.

Where does a competitor win instead of us, and what reason did the engine give? The reason is often the fix.

Which heavily cited pages don't mention us? That's your outreach list.

Which of our claims do engines actually repeat? The ones they skip are the ones nobody else says.

Which kind of question do we lose on? "Best X" losses and "how do I fix Y" losses need different fixes.

What do we fix first, and which answers should that change? Then check after the next run.

1. "What changed since the last run, and on which engine?"

Scores move. A headline drop feels like a verdict on your whole brand. Often it isn't.

Ask for the history, then the per-engine breakdown for the latest run. If one engine moved and the others held, the cause is probably on that engine's side: a model update, or a change in what it retrieves. Your content did not get worse overnight on one engine and nowhere else.

What to do with it: don't rewrite your homepage because one engine sneezed. Watch it for another run. If several engines move together, that's a real signal about you.

2. "On which prompts does a competitor get recommended instead of us? Quote what the engine said."

This is the most useful question on the list, and the word that matters is quote.

Engines usually give a reason when they recommend something: "best for teams that need X", "the usual choice if you already use Y". That reason is the engine telling you what it thinks your category is about.

Ask for the answers where you weren't mentioned, the competitors named in them, and the exact wording.

What to do with it: sort the reasons into two piles.

A reason that fits you too. You have the feature, but the engine doesn't know. That's a findability problem. The fix is getting that fact written down in places engines read.

A reason that doesn't fit you. You lose because of something you don't do, or don't do yet. That's positioning or product, and no amount of content fixes it.

3. "Which of the most-cited pages don't mention us?"

Engines lean on a small set of pages over and over. A comparison roundup cited 30 times that never names you is a page quietly voting against you on every one of those answers.

Ask for the most-cited sources, filtered to pages that don't mention your brand. Then ask your app to group them by type: review sites, editorial roundups, community threads, docs.

What to do with it: that grouped list is your outreach plan. Roundups get pitched. Review sites get reviews. Community threads get a real answer from someone on your team, labelled as such.

One limit to know: citations are pooled across engines, so this can't tell you which pages one particular engine prefers. We left that filter out on purpose, because the data behind it would mislead you.

4. "Which of our value propositions do engines actually recognise?"

You have a story about why you're better. Engines have their own version, assembled from whatever third parties wrote about you.

Ask which of your value propositions engines recognise, and which they don't.

What to do with it: the ones engines repeat are the ones other people repeat. The ones they skip are usually claims that only appear on your own site. Engines rarely take a company's word for itself: in our citation study, 2 of 3,543 citations pointed at a brand's own page when recommending it. Get the missing claims said by someone else: a customer, a reviewer, a benchmark, a docs page others link to.

5. "Are we losing on 'best X' questions or on 'how do I fix Y' questions?"

Not all prompts are the same kind. Renown groups them:

Discovery: "What are the best tools for X?"

Commercial: "X vs Y", "Is X worth it?"

Problem: "How do I stop Z from happening?"

Ask your app to count where you're mentioned and where you're not, split by type.

What to do with it:

Losing on discovery means engines don't file you under your own category. Work on category-level coverage: lists, roundups, comparisons.

Losing on problem means engines know you exist but don't connect you to the pain. Your content describes features. It needs to describe the problems they solve, in the words people use when they have them.

6. "What should we fix first? For each action, show me the answers it should change."

Renown already ranks your recommended actions. The better question makes your app tie each one to evidence from questions 2 to 5: which prompts and answers it should move.

An action with no answer attached is a guess. An action tied to three prompts where a competitor wins for a reason that also fits you is a plan.

What to do with it: do the top one. Mark it done from the same chat. Then, after the next run, go back to question 1 and see whether those answers moved. Checking is the part that makes this worth doing.

A caution about your AI app

Your AI app will summarise confidently whether or not it read everything. Two habits keep it honest:

Ask it to quote. A summary with the engine's actual words attached is checkable. One without them isn't.

Ask for counts. "Most answers" should come with a number, like 14 of 20.

The answers it reads are other AI models talking. The connector labels them as data rather than instructions, and they deserve the same scepticism you'd give any single source.

FAQ

What should I ask my AI visibility data first?

Start with where a competitor gets recommended instead of you, with the engine's exact wording. The reason an engine gives for its recommendation usually shows whether you have a findability problem or a positioning problem.

Why did my AI visibility score drop?

Check which engine moved. If only one did, the likely cause is a change on that engine, such as a model update or different retrieval, rather than something you did. Several engines moving together is a stronger signal.

Can I ask these questions without the connector?

Yes. The data is in the Renown dashboard. The connector lets an AI app like Claude or ChatGPT do the reading and counting, which matters most for questions 2 and 5 where the answer is buried across many responses.

Which cited pages matter most for AI visibility?

The ones cited most often that don't mention you. Each is a source engines rely on repeatedly, and each one currently tells them about your competitors instead.