Does the number of reviews on a review site relate to how often AI cites it?
A little, but weakly. G2 Learn (Kevin Indig, 2025-10-23, published on G2's own blog) took 30,000 AI citations from Profound and compared them with review counts across 500 software categories on G2. Result: categories with 10% more reviews had about 2% more citations; review count explained less than 2% of the variation (R² 0.009).
The author states the limits himself: it is a cross-section at one point in time, so it cannot tell whether reviews bring citations, whether heavily cited products simply get more reviews, or whether brand size drives both; it does not control for brand authority or content quality; and it studies only G2. The article also does not say which AI engines the Profound data covers. So the only safe reading is that review count is not the deciding factor.
Which review sites do AI answers actually cite?
AuthorityTech (Jaxon Parrott, 2026-11-03) used a monitoring question library between 2026-05-10 and 2026-11-03 and observed 8,343 ChatGPT runs, 8,461 Perplexity runs and 2,081 Google AI Mode runs, counting the share of runs that cited each review site:
| Review site | ChatGPT | Perplexity | Google AI Mode |
|---|---|---|---|
| G2 | 0.77% | 4.90% | 4.18% |
| Capterra | 0.18% | 1.78% | 0.43% |
| Software Advice | 0.16% | 0.99% | 0.14% |
| GetApp | 0.20% | 0.37% | 0.05% |
| Trustpilot | 0.10% | 0.02% | 0.00% |
Four things to know before reading this table:
- The article calls itself a first-party dataset, not an independent audit, and its publisher sells AI visibility services.
- The questions come from a monitoring library, not real user input; the ChatGPT numbers come from the API with web search, not the signed-in app.
- The ten review sites were picked by the authors and the list is not exhaustive; the numbers show a site was cited, not that its reviews made AI recommend anyone.
- The article says Trustpilot's own report reaches a different conclusion; we did not open that report, so it is unverified. The Trustpilot picture is therefore disputed and we do not judge it.
Even taking the table at face value, the highest number is only 4.90%: most answers do not cite a review site. A review site looks like one source in a minority of answers, not a gateway.
Do Product Hunt reviews help?
In a case study by Product Hunt team member Andrew Stewart (the page says "9 months ago" with no exact date and mentions changes in January, so probably early 2026) there is one concrete number: in ChatGPT answers comparing voice-to-text tools, Product Hunt pages mentioning Wispr Flow were cited in 6.7% of answers, and the author estimates Product Hunt contributed about 32% of Wispr Flow's ChatGPT visibility.
Three limits: it was written by someone at Product Hunt; it is an estimate, not a measured cause; and it does not count "reviews" separately, it talks about product and category pages. It also notes that Perplexity was blocked by accident and Copilot almost never cited the page, so the effect differs by AI product.
What did our own sampling show?
On 2026-11-06 we ran one question each in a search engine and looked at which review sites appeared. This is a single sample of search results, not AI answers, and must not be read as an AI citation rate.
| Question | Review sites in the results | Other results |
|---|---|---|
| best AI presentation maker reviews | A G2 category page, plus SlidesAI pages on Capterra and Software Advice | "Best of" lists from products and media |
| best AI sticker maker app reviews | One Stickermaker review page on G2 | List articles, app directories |
Neither Trustpilot nor Product Hunt appeared in either search. How to read it: whether a review site has a page for you depends on your category, and list articles far outnumber review-site pages, which fits what how to get into “best tools” articles says.
Which review site should an indie developer start with?
The order below is our inference from the material above, not a research finding. The rule: first see where your buyers compare products, then check whether that site has your category.
| Platform | Best for | What the evidence is | What to do first |
|---|---|---|---|
| G2 | Software sold to companies | Highest citation rate of the three sources, but the data comes from G2's own blog and a company that sells related services | Claim the product page, state category, price and features clearly; ask real users to write reviews |
| Capterra | Software sold to companies | Lower citation rate than G2, with a visible figure on Perplexity (1.78%) | You can ask the same real users to write there too; keep product details consistent with G2 |
| Product Hunt | Newly launched consumer and developer tools | One official case, an estimate for one category, not independent research | State clearly what the product is when you launch; then reply to comments and update the product page |
| Trustpilot | Consumer services, e-commerce | Almost absent in one count, the other is unverified, the disagreement is unresolved | Do it if you need the reputation it shows to people anyway; do not do it just for AI |
How exactly do you go about review sites?
1. Check whether your category exists
Search your category name on G2 and Capterra. It is worth joining only if there is a category page with similar products; without a matching category, no number of reviews gets you onto a list.
2. Unify the product description in one sentence
Use the same sentence in each review-site profile, your homepage and your About page: what it is, who it is for, what one thing it does. If your name collides with a bigger brand, write the distinction in.
3. Only ask real users for reviews
What a review says is up to the user. Do not buy reviews or write them from several accounts yourself; platforms act on it, and no study says it helps.
4. Reply to every review
Write the questions that keep recurring in reviews back onto your own pages; see How to write an "X vs Y" comparison page.
5. Test again with the same questions after a few weeks
Count how often you appear, not a single answer. For mentions versus backlinks see Do brand mentions matter more than backlinks for AI search; for the role of forums see Why AI answers like to cite Reddit.
FAQ
Do more G2 reviews make AI recommend me more?
In the analysis on G2's blog (Kevin Indig, 2025-10-23, 30,000 citations), categories with 10% more reviews had about 2% more citations, and review count explained less than 2% of the variation; the author says this does not prove cause.
Which review site do AI answers cite most?
In AuthorityTech's 2026-11-03 count, G2 is highest: ChatGPT 0.77%, Perplexity 4.90%, Google AI Mode 4.18%. It is first-party data and the publisher sells AI visibility services.
Does AI cite Trustpilot reviews?
It barely appears in that count (ChatGPT 0.10%, Perplexity 0.02%); the article says Trustpilot's own report concludes differently, and we did not open that report, so it is unverified.
Does Product Hunt help?
There is one case from the Product Hunt team itself: Wispr Flow's Product Hunt pages were cited in 6.7% of ChatGPT answers. It is an estimate, not independent research.
Can I ask friends or buy reviews?
Not recommended. Reviews should come from real users; buying them breaks platform rules and no study shows it works.
Sources
- G2 Learn · Do More G2 Reviews Mean More AI Visibility? Insights from 30k Citations （Kevin Indig, published 2025-10-23, checked 2026-11-06）
- AuthorityTech · Which Review Sites Does ChatGPT Cite? G2 Leads, and Trustpilot Barely Appears （Jaxon Parrott, published 2026-11-03, checked 2026-11-06, first-party data）
- Product Hunt · Case study: How Product Hunt can improve AI visibility in 2026 （Andrew Stewart, no exact date on the page, checked 2026-11-06）
- kagentx evaluations: the data on the case study pages; each example gives its date.