Who Google's AI Overview Cites, by Industry (We Checked 30 Buyer Searches)

By Minel Gunesoglu, founder of Is My Brand in AI · Original research, June 2026

TL;DR: We ran 30 buyer-style "best [category]" searches through Google's AI Overview, split across six industries, and logged every source it cited.

• Professional review media was the one near-constant, present in roughly 26 of the 27 AI Overviews that triggered.
• Reddit was powerful but concentrated. It appeared in about 41% of the answers, clustered in certain question types, and was completely absent from electronics.
• Brand-owned pages were rarely the cited source. The AI validates a brand through third parties, not through its own homepage.
• The mix changes by vertical. Finance, electronics, software, home, health, and auto each have a different citation pattern.

A few days ago we published a smaller study that found one thing: across 24 buyer searches, Reddit was the most-cited source in Google's AI Overviews. That held for the consumer-heavy sample we ran, and it was a useful headline. But it left an obvious question open. Does the same source dominate when you sell accounting software, or car insurance, or a camera? Those are very different worlds.

So this time we went deeper rather than wider in the same direction. We took 30 buyer "best [category]" searches and split them on purpose into six industries: software, consumer electronics, home and appliances, finance and insurance, health and personal care, and auto and travel. Five searches per industry. Then we read the AI Overview source panel for each one and wrote down every site it cited.

The fuller picture is more interesting than a single winner. There is no one source that wins everywhere. The closest thing to a universal answer is professional review media, which showed up in almost every AI Overview we saw. Reddit's strength is real, but it is concentrated in specific kinds of questions, and in one whole industry it never appeared at all. The practical takeaway is not "go do Reddit." It is "find out what your own category's AI answers are actually built from, because it is probably not what you assume."

How we ran the test

The method is the same reproducible recipe as the first study, so you can repeat it on your own category.

• Engine: Google AI Overview, the source panel it labels "N sites."
• Queries: 30 buyer-intent "best [category]" searches, chosen by real US search volume and spread evenly across six industries, five searches each.
• What we logged: for every search, each distinct domain in the AI Overview source panel, the total source count, whether an AI Overview appeared at all, and the type of each site (forum, review media, vendor, niche blog, video, news).
• When: a single snapshot, June 11 to 12, 2026, US English results.

Of the 30 searches, 27 produced an AI Overview and 3 returned a Shopping grid instead, with no AI answer. When an AI Overview did appear, it cited about 6 sources on average, ranging from 2 to 10. Every number below comes from that collection. Nothing here is estimated or filled in.

The citation map by industry

This table is the centerpiece. It is the same study viewed one industry at a time, because that is the level at which the pattern actually becomes useful.

Industry	Sources the AI Overview leans on	Reddit cited?	What stands out
Finance & insurance	NerdWallet, Bankrate, Forbes, Investopedia, U.S. News, CNBC	Only for "which company" questions (insurance, savings)	Banks' and insurers' own pages appear (GEICO, Allstate) but never on their own
Consumer electronics	RTINGS, PCMag, CNET, Forbes, ZDNET; niche blogs for cameras	Never (0 of 5)	Cameras were pulled from small enthusiast blogs, not big media
Home & appliances	RTINGS, Consumer Reports, Serious Eats, plus vendor sites	For durability and enthusiast picks (vacuum, fridge)	Pure commodities (mattress, air purifier) returned Shopping, no AI Overview
SaaS & software	Vendor blogs (Wrike, Zapier, Monday) plus PCMag and TechRadar	For "which tool" and small-business picks	Heavy reliance on vendor blog content
Health & personal care	Forbes, Fortune, Health Central, U.S. News, plus Amazon	Rare (multivitamin only)	Money-and-health topics lean on experts and niche health blogs
Auto & travel	Edmunds, Kelley Blue Book, Consumer Reports, U.S. News; travel media	For purchase advice (SUVs, EVs) and experience (airlines)	Airlines were Reddit-dominant; destinations were travel editorial

Read across the table and the lesson is plain. A finance buyer and an electronics buyer get answers built from almost entirely different sources. If you copy a playbook written for the wrong industry, you will spend effort in places Google's AI never looks for your category.

Lesson 1: Review and authority media is the real backbone

If you want one rule that travels across every vertical, this is it. Professional review and authority media appeared in roughly 26 of the 27 AI Overviews. RTINGS and PCMag for gadgets, NerdWallet and Bankrate for finance, Consumer Reports and Edmunds for cars, Serious Eats for kitchen gear, Condé Nast Traveler for destinations. Different names per industry, same role everywhere: the trusted reviewer is the source the AI reaches for first.

This is the most reliable lever in the whole study, and it is also the least glamorous. There is no shortcut. Getting covered or ranked by the review media your category already trusts is slow relationship and reputation work. But it is the one move that pays off in every vertical we tested, including the ones where Reddit was nowhere to be found.

Lesson 2: Reddit is powerful, but concentrated

Our first study's headline still stands for the sample it was drawn from. Reddit really is a heavy hitter. Across this larger set it appeared in 11 of the 27 AI Overviews, about 41%, and in one search ("best airlines for long flights") the top three cited sources were all Reddit threads. When real lived experience is what the buyer wants, the AI goes straight to the forum.

What the wider sample adds is the boundary. Reddit shows up for a specific family of questions: experiential picks, "which one should I actually get," durability and "buy it for life" questions, and small-business tool choices. It clusters in airlines, SUVs, electric cars, robot vacuums, refrigerators, insurance, savings accounts, project management, and CRM.

It is absent for a different family: spec sheets, rate tables, and pure professional-review questions. Reddit appeared in zero of the five electronics searches. It stayed out of credit cards, CD rates, coffee makers, running shoes, travel destinations, and "best cars." So the honest, fuller story is not that the first study was wrong. It is that Reddit's dominance is concentrated, not constant, and you need to know which side of that line your category sits on before you invest in it.

Lesson 3: Your own site is rarely the cited source

Vendor and brand pages did appear, in about 17 of the 27 AI Overviews, so this is not "your website is invisible." Proton, Wrike, Monday, Bose, Samsung, GEICO, Allstate, and Amazon listings all turned up. But they almost never appeared as the lead source, and essentially never as the only source. They sat next to a reviewer or a Reddit thread, included as one voice in a wider answer the AI assembled from third parties.

This is the core idea behind everything we build. The AI does not take your word for your product. It validates you through other people: the reviewers who rated you and the forums where buyers discuss you. Your own page matters, but as the thing being talked about, not as the source doing the talking. To be cited, you have to be present where the AI looks, not only on the page you control.

A note on commodities

Three of the 30 searches produced no AI Overview at all. "Best seo tools," "best mattresses," and "best air purifier" returned a Shopping grid of products instead of a text answer. The pattern is consistent: for pure physical commodities, Google often skips the AI Overview entirely and routes the buyer to a product carousel.

If you sell something in that bucket, the AI-citation game is partly beside the point. There is no AI answer to be cited in. The battle moves to Shopping feed and product-listing optimization, plus the reviews that feed it. Worth checking your own category before you assume an AI Overview is even the surface you are competing for.

The playbook: how to use this in your category

Here is the part you can act on tomorrow. Five steps, in order.

1. Find your vertical's pattern first. Run your own "best [your category]" search, open the AI Overview source panel, and read the cited sources. Do not assume Reddit, and do not assume review media either. Look at what is actually there for your industry.
2. Win the review and authority media of your category. This is the one constant across all six verticals. Identify the few sites the AI already trusts for your space (RTINGS for gadgets, NerdWallet for finance, Edmunds for cars) and earn coverage or ranking on them. Slow, but it works everywhere.
3. Use Reddit where it actually appears. If your category is experiential, durability-driven, "which one should I get," or a small-business tool choice, the relevant subreddit is high-leverage. Be genuinely useful there. If your category is spec-driven or YMYL finance, spend that energy elsewhere.
4. Stop expecting your own page to be the source. Keep your page clear and easy to quote, but measure success by your third-party citations: the reviews and forum threads that mention you. That is what the AI is reading.
5. If you sell a commodity, check whether an AI Overview even fires. If your search returns Shopping instead, focus on product-feed and listing optimization rather than chasing a citation that does not exist.

If you want the surface-by-surface version of this, see AI Overview optimization and how to get cited by ChatGPT. And before you act, it helps to know where your brand actually stands today. Our guide to tracking brand mentions shows a free way to check by hand, and we are building a free AI visibility checker (early access) to make that a one-click job.

Honest limitations

This is a snapshot taken carefully, not a law of physics, and the caveats matter.

• It is a single engine (Google AI Overview), a single market (US English), and a single two-day snapshot. AI Overviews are personalized and volatile, so another day or account can return a different mix, and whether an AI Overview appears at all can change.
• 30 searches is a defined, volume-ranked sample of "best [category]" buyer intent, five per industry. It is enough to show per-industry patterns, not enough to rank the entire web. We frame Reddit's presence as a pattern across the sample, never off a single query.
• We classified each source by its title and brand, and counted by whether a domain appeared in a given AI Overview, not by how prominently it was placed. Grouped citations were counted by their lead domain.
• Industry buckets hold different raw search volumes, so the comparisons here are about the pattern within each vertical, not the size of one vertical against another.

Taken together with the first study, the direction is clear. The earlier work found that AI Overviews lean on third parties over brand-owned pages. This one shows which third parties, and that the answer depends on your industry. Review media is the constant. Reddit is the powerful variable. Your own homepage is rarely the source.

Method, full query list, and per-search results were logged during collection. If you are a journalist or researcher and want the underlying notes, reach us through the contact page.