AI Overview audit: who Google cites in your industry

The most common symptom. Your page is still in the results, but the answer above them closes the question and nobody scrolls further down.

Queries starting with “how”, “what is” and “how much does it cost” lose traffic faster than buying queries, because those are where Google most often answers on its own.

Google gives three sources in the answer and none of them is yours, even though you rank above them in the list of links.

An AI Overview audit is a check of which queries in your industry make Google generate a model-based answer, which pages it cites in it and how those pages differ from yours. It is a snapshot, not an ongoing service: it ends with a list of queries, a list of fixes and a measurement date.

of Google AI Overviews answers to seven buying questions about agencies mention JustIdea, one in four

of model answers to those questions mentioned JustIdea in the first month of measurement, across five engines combined

Three doubts we hear before an AI Overview audit

„We can check it ourselves, just type it into Google”

It is worth checking to see the scale. The problem starts with comparison: the answer differs between sessions and locations, so a single check says nothing about how often you are missing from it.

„It will change again in six months anyway”

It will, and that is why the report is dated rather than a document for years. The fixes that come out of it are content format and structured data, things that work regardless of what Google calls its product.

„We would rather go straight to implementation, why audit”

Because without it you do not know how many of your queries get an AI answer today. In some industries it is one in five, in others more than half, and that decides whether it is worth spending money here.

There is only one order: without a readable site there is nothing to cite, so if both areas are neglected, we start with a classic SEO audit. This audit is a one-off product: it has an end date, a list of queries and a list of fixes, and you decide whether to implement them in-house or hand them on.

Anyone who promises a place in AI Overviews promises something they do not control: Google's documentation has no tag or setting that gets you into those answers. An AI answer also does not appear on every query and it changes over time, which is why we measure the same set of queries repeatedly and record the date of each measurement.

• You rank decently in Google, yet you do not see yourself in AI answers.
• You want to know whom Google cites on queries in your industry.
• You need a list of fixes with an end date, not a subscription.
• You are considering ongoing work and want to check first whether it makes sense.

• The site has technical problems or weak rankings: start with an SEO audit.
• You want someone to implement fixes every month: that is ongoing work.
• You are asking about one model, not Google: ChatGPT has its own page.

Artificial intelligence in the search engine does not pick pages the way a classic ranking does. A generative algorithm assembles the answer from fragments of several sources at once and only then lists the links beneath it. So what matters is not a high position alone, but whether your fragment can be lifted out and placed into the sentence the model is building. A page can sit at the top and never be used once, because the answer to the question is buried three paragraphs deep.

This translates directly into organic traffic, and that is why the audit starts with measurement, not recommendations. When the answer sits above the results, some users no longer scroll down: they get what they came for without visiting a site. If your brand is in that answer, you lose the click but gain the mention. If it is not, you lose both. That difference can be measured, and we show it on your set of queries.

We treat this as an investment with a measurable return, not a novelty. You pay for the audit once, you get a list of fixes ordered by effort and expected effect, and once it is implemented the same measurement is repeated on the same queries. We do not push an ongoing package either: if the first measurement shows that your industry does not yet get AI-generated answers, we will say so plainly and put the topic off for a quarter.

AI visibility audit: questions before you decide

AI visibility audit: four things we say at the start