Best AI Overviews Tracking Software: What a Local Business Needs

Google Search Console's Generative AI performance report is free and shows impressions from AI features, and paid trackers such as Otterly or Ahrefs Brand Radar run your customer questions on a schedule. Start with a manual check of 20 real customer questions. Pay for a tracker only if that check shows a pattern worth watching monthly.

The reason to start by hand is that a tracker is only as good as the questions you give it, and writing those questions is the real work. A small business that spends an hour on the spot check below will know whether AI Overviews are even a problem for them, and will arrive at any vendor demo knowing exactly what to measure.

Follow me on Instagram@sagnikteaches

The 20-question spot check, step by step

The aim is to find out, with a few hours of your own time, whether Google's AI answers name your business, name a competitor, or skip your sector entirely. Use a private browsing window so that your history and logins do not shape the answers.

Connect on LinkedInSagnik Bhattacharya

1. Write 20 questions in customers' words. Take them from your enquiry emails, the phone calls you answer, your reviews and the questions staff hear. Real phrasing beats keywords: "how much does it cost to move a two-bed flat" rather than "removals pricing".
2. Spread them across five types. Four each of: a service question ("who does X"), a price question, a comparison ("X or Y"), a trust question ("is X reliable") and a how-to or "what should I know" question.
3. Search each one twice. Do one search now and one a day later, ideally on a different device. Note whether an AI Overview appears at all.
4. Log what you see. For each question record: AI Overview shown (yes/no), your business named (yes/no), your website cited as a source (yes/no), competitors named, and any wrong fact about you.
5. Add a rating for the wrong facts. An AI Overview that gives the wrong opening hours or a service you do not offer matters more than one that omits you.

Here is an illustration. Say a small removals firm logs its 20 questions in a sheet. The result: an AI Overview appeared for 11 of the 20 (55%). The firm was named in 2 of those 11 and its website was cited in 3. A competitor was named in 7. One overview quoted a price range for a one-bed move that is well below what the firm charges, taken from a comparison site. That is a very useful picture: the firm has a visibility gap on about nine questions, a price-expectation problem on one, and a clear list of pages to improve. It also shows that nine questions had no AI Overview at all, so there is nothing to track there.

Subscribe on YouTube@sagnikteaches

Five rows are enough to show the shape; your own sheet will have twenty. Keep the log somewhere you can copy it forward, because next month's check should use the same questions so that you can compare like with like.

Use AI to draft the question list, then edit it

If you are short on ideas, ask a chat assistant to propose questions, then throw away half of them. Give it your real material, not an empty request:

I run a small removals firm. Below are 12 customer enquiries and
6 reviews. List 30 questions a customer might type into Google
before choosing a firm. Use their wording, not marketing wording.
Group them as: service, price, comparison, trust, how-to.
Mark any question that mentions a place so I can remove it.

[paste enquiries and reviews here]

A typical result includes a few stiff questions ("What are the benefits of professional relocation services?") that no customer would type. Delete those and keep the ones that sound like your inbox. The 20 survivors are your tracking list for the free checks and for any paid tool you try later.

What the free reports already show

Two free sources tell you something without any new software, and you should look at both before paying for anything.

Search Console. Google announced a Generative AI performance report in June 2026 and has been rolling it out; a site with very few AI impressions may not see one yet. It reports impressions only, and it combines AI Overviews and AI Mode in one figure. That means it cannot tell you which question produced the impression, whether you were named in the text or only linked, or what the answer said. It does show whether the trend is up or down, which is worth a look once a month. Google has also said there is a separate control to opt a site out of generative AI features, but inclusion is the default and is not a ranking signal.

Bing Webmaster Tools. Microsoft's AI Performance report, in public preview since February 2026, shows citations and the questions behind them. It covers Microsoft's own AI answers rather than Google's, but it is the closest thing a small site has to a free citation report, and the questions it lists are a good source for your own list.

Neither report replaces the spot check, because neither shows what the answer actually said about you. For a deeper look at how AI Overviews change local results, the tutorial on how AI Overviews are changing local search sets out what is happening on the results page itself.

Why one screenshot proves very little

People often check one question, see a competitor in the AI Overview and panic, or see themselves and celebrate. Both conclusions are weak. AI answers can differ from person to person, device to device and day to day, and the same question phrased slightly differently can give a different answer or none at all. Google also does not show an AI Overview for every search.

Treat visibility as a rate across many questions, not a yes or no on one. In the removals example, "named in 2 of 11 AI Overviews" is a rate. If next month it is 4 of 12 after you improve three pages, that is a result you can act on. If it moves from 2 of 11 to 1 of 10, the movement may be noise. This is also the main thing paid trackers sell: the same questions, run the same way, on a schedule, so the noise averages out.

What paid trackers actually do

A tracker takes your list of questions (usually called prompts), runs each one on a schedule in a clean session, saves the answer and reports three things: whether your brand is mentioned, whether your site is cited, and which competitors and sources show up. Most also score sentiment and suggest content changes. What differs between products is which AI engines they cover, how many prompts you get, and how often they run.

Prices below are the list prices on each vendor's page as I checked them in October 2026, in USD. They move often, so confirm on the page before you decide.

Two points stand out. First, entry tiers are small. Otterly's $29 plan covers 15 prompts, so a 20-question list does not fit and would push you to the 100-prompt tier at $189. Second, the tools differ in whether they track Google's AI Overviews specifically or AI answers in general. Confirm that the plan you are looking at covers Google AI Overviews, not only ChatGPT and Perplexity.

Is it worth paying? A sum for a small business

Suppose a landscaping company has done the spot check and settled on 15 priority questions. The Otterly Lite plan, at $29 a month, would track them: $348 a year. Ask what that buys. It buys a trend line and an alert if a competitor suddenly dominates, and it does not buy any jobs. For the spend to make sense, the trend has to change what you do.

A rough test: if an average job brings in $400 of margin, $348 a year is less than one job. That sounds easy to justify, but the tracker cannot show you whether any job came from an AI Overview. The honest question is whether you would act differently with the data. If the answer is "I would rewrite the patio page when the number drops", it earns its fee. If the answer is "I would just look at it", it does not.

Now take the other end. A tutoring centre with six locations (so, many more questions across many places) wants 100 prompts and would pay $189 a month, about $2,268 a year. That only makes sense if someone owns the data and acts on it each month. Without that owner, the subscription turns into a dashboard nobody opens.

Reading a tracker's report without being fooled

If you do try a tracker, most of its first report will look like this illustrative example for a landscaping company tracking 15 questions over a week:

Brand mentioned: 4 of 15 prompts (27%)
Website cited:   2 of 15 prompts (13%)
Top competitor:  mentioned in 9 of 15 prompts (60%)
Most cited sources: two comparison sites, a trade directory,
                    the competitor's blog
Sentiment: neutral

Three checks stop you reading too much into it. First, open the raw answers behind the score for at least five prompts, because a "mention" can be a list of ten firms in which you are the last. Second, look at which sources are cited. If two comparison sites supply most of the citations, your fix may be getting an accurate listing there, not rewriting your own pages. Third, compare the tracker's answer with what you see yourself in a private window. If they disagree on several questions, ask the vendor how it runs the searches and whether it uses a location or account setting that differs from your customers'.

Before paying, ask the vendor four things: how often each prompt is run, whether Google AI Overviews are tracked as a separate engine, whether you can export your raw answers, and what happens to your data and prompts if you cancel. A vendor who cannot answer the second question clearly is probably tracking chat assistants, not Google.

A decision rule for a local business

• Do the spot check and use Search Console first. It costs nothing and answers whether AI Overviews appear for your questions at all.
• If fewer than a third of your questions show an AI Overview, you do not need a tracker. Repeat the check each quarter.
• If many do and you are rarely named, fix the pages first (service pages, prices, FAQs, business profile details), then track to see whether it worked.
• If you are named often and want to protect it, or you have several competitors worth watching, a small paid plan can make sense.
• If an agency offers to sell you tracking, ask who owns the prompts and the data. The tutorial on choosing an SEO agency that understands AI search has questions for that conversation.

When an AI Overview gets a fact about you wrong

The removals firm from the spot check found a price range far below its own. Tracking cannot fix that, but the log tells you where to look. Click through the sources the overview cites and find the page the figure came from. It is usually an out-of-date listing, a comparison site or an old page of your own. Correct what you control, ask the third party to update what it holds, and state your real price range plainly on your own pricing page, with the date it was last reviewed. Then repeat the search after a few weeks and note whether the answer changed. Google also provides a feedback link on its AI answers, which is worth using for a clearly wrong claim, though it gives no guarantee of a change.

What no tracker can tell you

It helps to know the limits so that you do not pay for certainty you will not get. A tracker cannot tell you how many customers saw an answer, what share of your website visits came from it, or why a particular answer changed. Search Console reports impressions, not the text of answers. Google's own guidance is that AI Overviews and AI Mode need no special optimisation beyond normal search practice, and that no new AI-specific files or special markup are needed to appear. If a vendor says a particular file or tag will win you a place, ask for evidence you can check.

The tracker also sees only the questions you give it. If your real customers ask something you did not list, you will not see it. Refresh the list every quarter from new enquiries and reviews.

A realistic mistake and how it showed up

Consider a bookshop that tracks the question "best independent bookshop" and sees itself named every week. It reports success, but its sales do not move. A look at the enquiries shows what customers actually ask: "do you stock [a specific title]", "do you run a book club", "can I order a book for collection". The tracked question was one the shop could win and nobody was typing. Moving to the real questions showed the shop appeared in none of the AI answers for stock queries, because its catalogue was not on the website. The fix was a page, not a subscription.

A 30-minute monthly routine without paying

1. Open your 20-question sheet and repeat the searches in a private window (about 15 minutes).
2. Update the counts: AI Overviews shown, named, cited, wrong facts.
3. Check the Search Console report for the month's trend (5 minutes).
4. Pick the one question where a competitor is named and you are not, and write down the page that should answer it.
5. Improve that single page, and note the date so you can see whether the next check changes (10 minutes of planning, more of writing).

This is the same discipline a paid tool automates, with the advantage that you read the actual answers rather than a score. For a broader view of what to review on your site at the same time, the tutorial on auditing your website's SEO with AI in one afternoon gives a one-session method, and what AI can and cannot do for local SEO sets realistic expectations about the effort involved.

Whatever you choose, write down your starting figures today. A tracker's first dashboard is only useful if you know what you saw before it. If you want help building the question list or deciding whether a tracker is worth it for your business, that is the kind of problem my AI implementation consultation covers, and you can get in touch here.

Questions about tracking AI Overviews

Does Google Search Console show AI Overviews?

Search Console has a Generative AI performance report that shows impressions from AI features. As of October 2026 it combines AI Overviews and AI Mode in one figure and reports impressions only, so it cannot split the two or show which exact answer named you. It is free and a sensible first look.

Why do I see different AI Overviews from my customers?

Answers can vary by person, device, time and wording, and an AI Overview does not appear for every search. A single check on your own signed-in browser can mislead. Check the same question more than once, in a private window, and treat a pattern across 20 questions as the result, not one screenshot.

How many questions should a local business track?

Twenty is enough to start. Choose the questions customers genuinely ask before they book, buy or call, split across services, prices, comparisons and trust. More than about 50 rarely tells a small business anything extra, and paid tools charge by the number of questions tracked.

Can I improve my AI Overviews visibility directly?

Google says AI Overviews need no special optimisation beyond normal good search practice, and that you do not need to create new AI text files or special markup to appear. Clear pages, accurate business details, reviews and content that answers real questions are the levers. A tracker measures the result; it does not change it.

Is a free spot check really enough?

For many small businesses, yes. If a monthly 30-minute check shows you named for most of your key questions, or shows that no tool could fix a basic page problem, paying adds little. A tracker earns its fee when you need a trend over many questions or several competitors.

Further reads

• How Wedding Businesses Get Found When Couples Ask AI for Suppliers — A worked example of getting named when people ask AI for suppliers.
• AI SEO for Plumbers: Get Named by ChatGPT and Google AI Overviews — See the same visibility questions applied to a trade business.
• Best AI SEO Agency for Hotels? Questions to Ask Before You Sign — Questions to put to an agency that promises AI search results.
• How to Write Website Articles With AI That Still Rank on Google — Write pages that deserve to be quoted, not just indexed.
• What Are the Best AI Visibility Products for a Local Business? — Which AI visibility tools suit a single-site local business, what each one tracks, and a free manual check to run first.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Otterly, Ahrefs (Brand Radar), Profound, Peec AI and SE Ranking pricing and product pages, and Google Search Central guidance on AI features, all checked in October 2026.