No review profile? A 1% AI citation rate
A study of 804,491 AI answers sorted 1,926 brands by how developed their Trustpilot profile was. Brands with no profile turned up alongside a Trustpilot citation in about 1% of the answers measured. Brands on the next rung up, which starts at as few as 1 to 13 reviews, turned up in 53.5%. Before you act on a gap that size, you should know that Trustpilot paid for the research and that the thing being measured is a good deal narrower than the headlines made it sound. Here is what survives a careful reading, and what a small business should do about it.
A minimal profile moves the number 52 points
Seer Interactive published the research on 14 May 2026, two days after Trustpilot's own press release. The method is disclosed more openly than most vendor studies manage - 804,491 AI responses across 1,926 brands on four AI platforms, drawn from 15,783 prompts over a window in March 2026, across eight verticals. Brands were sorted into four tiers by profile state: none verified, minimal, active, and actively managed at high volume.
The gap between the first two tiers is the whole story. Tier 0 brands had a median citation rate of 1%, while Tier 1 brands sat at 53.5%, and Tier 1 starts as low as 1 to 13 reviews. Trustpilot's own release puts the next tier at 75.3%, reached when a brand collects over 80 reviews and responds regularly. Above that, in a stricter cohort matched for domain rating, fully optimised profiles beat active-but-not-optimised ones by a further six points.
The number that survives the hardest scrutiny is a quieter one, further down the study: the domain-rating comparison. A brand with a Domain Rating of 0 to 15 and a minimal profile reached 54.5%, near-identical to a mid-authority brand at DR 16 to 35 on the same tier at 52.6%. That is the closest the study gets to separating the profile from everything else a well-funded marketing department does, and it says the ladder is climbable without authority you have not got.
What this study is, and what it is not
Trustpilot commissioned the research. The release states it plainly in a footnote, Trustpilot published first, and the company had shipped an AI search analytics product the month before. Seer is a serious research shop and the methodology is unusually well documented, but the funding belongs in the first paragraph of anything quoting the figures, including this post.
It is also correlational. Nothing was randomised, and brands that build review profiles are brands that invest in marketing generally. An independent write-up put the problem well: the reviews may correlate with citations without being the lever that moves them, and the study does not isolate reviews as the cause. Seer concedes as much in the article, noting that marketing spend and audience size may have influenced the rates.
Then there is the definition. Citation rate here means how often a brand appeared alongside a Trustpilot citation in answers to review-shaped buying-journey prompts. It is not general AI visibility, and it is not how often an assistant mentions your company. Seer describes the jump as the difference between being "functionally invisible in AI citations that mention Trustpilot" and being consistently present, which is a much narrower claim than the press made of it.
That press is a good lesson in how a figure degrades. The study resurfaced in August through a Forbes Councils contributor column, a fee-based membership slot rather than Forbes journalism, which reported 1% rising to 75% for a brand with "a review presence". A review presence gets you 53.5%; 75.3% needs 80-plus reviews and responses. Elsewhere it appeared as brands being "75 times more likely" to show up. Even Seer's own news page restated the 53.5% rate as "53% more likely". Read the primary source before repeating any version of it.
Finally, every hard number here is US-sourced. Seer reports no geographic split at all, the consumer surveys are US panels, and the independent citation samples are US queries. Whether Trustpilot, Google reviews, Yelp or a local directory carries the weight in Ireland, the UK or the Nordics is an open question, and the agency posts claiming to answer it have nothing published behind them.
The engines are not going looking for reviews
The most useful part of the study is the plumbing underneath it. 99.5% of those citations arrived by way of organic search rankings. No engine was reaching for a review platform deliberately. They ran their own sub-searches, review pages were already ranking for review-shaped queries, and those pages came back. Half a percent came from a fan-out query naming Trustpilot. That makes this a supply-chain fact rather than a review-platform pitch, the same shape we described in the AI citation supply chain you don't own: the surfaces that answer questions about you get ranked and quoted whether or not you have ever logged into them.
A brand with no third-party reviews does not simply get skipped. ChatGPT says so out loud, in phrasing Seer recorded as "[Brand] lacks reviews on trusted sites like Trustpilot, BBB", in several cases while citing the brand's own website in the same answer. The name is in the answer and the credibility is not, which is worse than being left out, because the reader now has a stated reason to hesitate.
Which prompts pull review surfaces in at all is the second finding worth keeping. Review and trust sites made up 1.51% of citations at awareness prompts, 16.24% at evaluation and 24.27% at intent, against a 14.01% aggregate. A general category question barely touches them. A direct "should I choose this company" question leans on them for a quarter of its evidence, so test the intent-shaped questions and ignore the category ones.
Per engine, controlling for brand size and organic strength, ChatGPT led at 57.9%, Perplexity followed at 51.9% and Google AI Mode came in at 48.9%. Google AI Mode had the most varied mix, reaching for Yelp and G2 more often than the others. Gemini cited review sites least and Seer published no figure for it, so treat any Gemini number you are quoted as invented.
Review sites own a narrow band of the answer layer
Read only the Trustpilot study and you will badly overestimate how much of the answer layer review platforms occupy. Ahrefs' September 2026 snapshot of Google AI Overviews, across more than three million US queries on every topic, ranks Trustpilot 35th at 0.4% of mentions and Yelp 49th at 0.3%, with Reddit second at 18.5% and YouTube first at 22.9%. G2 and Capterra do not make the top 50.
Both are accurate because they answer different questions. Seer prompted deliberately for review-shaped buying journeys in eight commercial verticals; Ahrefs sampled everything AI Overviews answers, most of which has nothing to do with picking a supplier. Review platforms matter enormously on a narrow band of high-intent questions and very little across the general web.
Category decides the platform, too. Profound's data, published by G2, has G2 supplying about 33% of cited reviews on ChatGPT and Google AI Overviews, and closer to 75% on Perplexity, within a basket defined as software review sites. "G2 dominates" and "Trustpilot dominates" are both true because they describe different shelves. Software buyers get G2 and Capterra, consumer services get Trustpilot and Google, hospitality gets Tripadvisor, which sits 18th in the Ahrefs table at 0.6%, ahead of Trustpilot. Some of that supply is now contractual: OpenAI licensed Yelp's reviews, ratings and photos into ChatGPT's local answers in July 2026.
Our own Irish audits show the same shape from the other end. When we followed every citation behind two Irish hotel questions, 32 came back across the AI answers and not one was a hotel's own website. Every source backing a named recommendation was a review platform, a listicle, a travel desk or a forum thread. And when we audited The Shelbourne's reputation surfaces, it held 4.5 stars or better everywhere it was correctly listed, on three of the ten platforms we checked.
One bad review, copied five times, reads as five opinions
A claimed profile is a surface you then have to maintain, and the mechanism that carries a good reputation carries a bad one just as far. Seer ran the experiment on itself in March 2026. In 24 years the agency had collected exactly one bad review, from 2018, syndicated across five directories. Tracking branded prompts, they found the phrase "high account manager turnover" appearing 67 times, with Clutch in 16% of branded prompt outputs and AgencySpotter at 6%. On three of those sites the only review listed was that same one.
The mechanism is the part to remember. As Seer put it, when an LLM sees the same claim on multiple sites, it does not flag it as potentially duplicated, it reads it as corroboration. One complaint, copied across five directories by the directories themselves, arrives at the model looking like five independent sources agreeing.
Their fix did not hold either. They published a page addressing the claim, it held for a few months, then it lost ground. They call it marketing whack-a-mole, which is a more honest description of reputation work than most agencies give.
What you actually do
If you have never claimed a third-party profile, this is the cheapest visibility work available to you: an afternoon rather than a project, and the domain-rating comparison says it works without the backlinks a bigger competitor spent years building. Four steps, in this order.
1. Find out what the engines already cite for your category. Before claiming anything, ask an assistant the intent-shaped questions a customer would ask about a business like yours, and read which domains the answer leans on. If they are Tripadvisor and Google, a Trustpilot profile is not your first move. If they are G2 and Capterra, Trustpilot is close to irrelevant.
2. Claim the two or three profiles that carry your category. Not all ten. Every extra surface is another thing to keep current, and a claimed profile with stale details and no reviews is a worse signal than a gap. Get the name, address, hours and description right, and check that the listing is you rather than a similarly named business, which is a failure we have found on real audits.
3. Fix the stale duplicates before you chase volume. Old listings you never created, syndicated copies of one complaint and abandoned profiles from a previous owner all read as separate corroborating sources. Search your own business name plus "reviews" and work through what comes back. People skip this step. It is the one that does the most damage when left.
4. Ask for reviews, then answer them. The 75.3% tier is defined by volume plus responses. Consumers want the same thing from the other side: BrightLocal's 2026 survey found 47% will not use a business with fewer than 20 reviews, and 74% only care about reviews written in the last three months. Recency is a standing requirement, so one annual push does not hold the position. The audience is growing fast too: consumers using AI to find local recommendations climbed from 6% in 2025 to 45% in 2026.
A reputation audit does the first three steps in one pass: which platforms carry your business, what rating each shows, which listings are wrong or belong to somebody else, and where the gaps sit.
Questions people ask about this
Sources: Seer Interactive, Trustpilot via PR Newswire, Ahrefs, BrightLocal, G2 citing Profound, Search Engine Land, BrandMentions and Forbes Agency Council, drawn from publicly available reports and studies published in 2026. The Seer research was commissioned by Trustpilot. Irish audit figures are from public Baseline Labs scans.
