G2 and Capterra show up in AI answers mainly on one engine. We checked this in our audit of 25 software-buying prompts across ChatGPT, Perplexity, and Google AI Overviews. Perplexity cited a review site in more than half its answers. ChatGPT and AI Overviews rarely did, and no engine cited a vendor's own review profile.
For G2 and ChatGPT specifically, the short answer is that ChatGPT rarely cites review sites at all. Keeping a complete profile is sensible hygiene, but don't expect review volume to move AI answers much.
Key Takeaways
- Review sites matter mostly on Perplexity. It cited one in 14 of 25 software-buying answers, while ChatGPT and Google AI Overviews did in 2 each.
- No engine cited a vendor's own review profile. They cited the sites' category pages, "best X" articles, and comparison pages.
- ChatGPT retrieved 9 product pricing pages on review sites but cited none of them.
- Keep your profile complete, but don't expect more reviews to do much. Kevin Indig's analysis for G2 ties review volume to under 2% of the variation in AI citations.
Which AI Engines Cite G2 and Capterra?
Perplexity is the engine that cites G2 and Capterra. In our audit of 25 software-buying prompts, it cited a review site in 14 of 25 answers. ChatGPT did so in 2 of 25, and Google AI Overviews in 2 of the 21 answers where an Overview appeared.
The prompts covered five categories: CRM, project management, help desk, HR and payroll, and developer monitoring. Each was asked five ways: plain "best X", company size, integration, price band, and "alternatives to" the leader. ChatGPT ran logged out from the US with search on. Perplexity ran through its Sonar API, and AI Overviews on US desktop Google. The wider results are in our guide to GEO for B2B SaaS.
Perplexity's 37 review-site citations leaned on two domains. G2 took 19 and Capterra 8, while SelectHub and GetApp had 3 each, Software Advice 2, and PeerSpot and SoftwareSuggest 1 each. The pattern was strongest on the broadest prompts: Perplexity cited a review site in all 5 plain "best X" answers. On the alternatives prompts it was 3 of 5, and on the size, integration, and price prompts 2 of 5 each.
What Does ChatGPT Retrieve From Review Sites?
Most review-site pages ChatGPT retrieved were pricing pages, and it cited none of them. Of the 1,056 search results it retrieved, 14 were review-site pages, or 1.3%, in 8 of 25 answers. Nine were product pricing pages on G2, TrustRadius, and Capterra, such as g2.com/products/gusto/pricing. The rest were 3 category or directory pages and 2 learn.g2.com articles.
Its searches rarely asked for reviews. Only 2 of its 90 fan-out searches, the queries it runs before answering, contained the word "reviews", and none named a review site. Price came up far more: 56 of its 90 fan-out searches contained a price word.
The citations went elsewhere. ChatGPT cited 34 pricing pages across our prompts, and 31 of them were on vendors' own sites. None was a review site's pricing page. Where a price mattered, the page it cited was usually the vendor's own.
One outside study found ChatGPT citing review sites even less. DerivateX ran one buyer-intent prompt in each of 40 software categories through ChatGPT with web search, and "G2 and Capterra received zero" citations. Review aggregators made up 1.1% of cited sources. It covers ChatGPT only, across 40 prompts, and counts cited links, not influence. DerivateX also sells AI-visibility services. Our ChatGPT answers point mostly the same way. Of 103 citations, two went to a review site, and both were learn.g2.com "best X" articles rather than profiles.
What Review-Site Pages Do AI Engines Cite?
AI engines cite review sites' own lists, not your profile. Across our prompts, the three engines made 41 review-site citations, and none was a single product's review page. They were mostly category and directory pages, the sites' editorial "best X" articles, and comparison or alternatives pages.
Perplexity accounts for 37 of those citations. Nineteen were category or directory pages, such as g2.com/categories/crm and capterra.com/help-desk-software/. Twelve were editorial articles, mostly "best X" lists on learn.g2.com and capterra.com/resources/. Five were comparison or alternatives pages, and one was a G2 discussion thread. The G2 pages under a product's name, like g2.com/products/jira/competitors/alternatives, were alternatives pages rather than reviews. The other engines showed the same pattern, on a much smaller scale. ChatGPT's two were both "best X" articles. AI Overviews' two were a G2 category page and a Gartner Peer Insights alternatives page.
On integration prompts, Perplexity cited directory pages filtered by integration, for example a GetApp page of CRM software integrations with Slack. Presumably, a page filtered that way only shows tools whose listing records the integration. An incomplete listing could keep you off it.
Why would an engine skip the profile? We think it's a matter of fit. Our prompts asked for options: the best help desk, say, or alternatives to a leader. A profile describes one product, while a category page or a "best X" article lists many in one place. That's the shape of answer the engine is building.
What gets cited, then, is the review site's framing of your category. That puts three things ahead of the profile page itself: your categories, your integration listings, and a place in the site's own editorial lists. In practice, that means checking which categories you sit in on G2 and Capterra before asking for more reviews. Those "best X" articles are lists too. Our guide to getting your SaaS into "best tools" lists covers that side.
Do Google AI Overviews Cite Review Sites?
In our audit, Google AI Overviews rarely cited review sites. Only 2 of 21 answers did: one cited a G2 category page and the other a Gartner Peer Insights alternatives page. The Overviews leaned on lists, Reddit, and YouTube instead. They cited Reddit in 12 of 21 answers and YouTube in 6.
Google showed an AI Overview for 21 of our 25 prompts, yet review sites made up just 1.4% of the Overviews' 148 citations. Reddit alone took 11.5%. Lists made up 58.8%, so the page an Overview cited was most often a list.
The regular results on the same pages looked similar. Across 214 organic results on the 25 result pages, review sites were 3.3%. Independent lists were 54.7%, and Reddit held a top-3 organic slot on 20 of 25 pages. Google didn't put review sites up front in its organic results for these prompts, and its Overviews didn't either.
Our Google sample is small. Four prompts drew no Overview, including "best help desk software" and "best HR and payroll software", so the base is 21 answers. A different set of prompts could easily shift a count that small.
SE Ranking's study points the other way. It checked 30,000 commercial keywords in the US, and 22,729 of them showed an AI Overview. Its finding: "Out of all AI Overview responses, 34.5% cited at least one review platform." Of the review-platform links, Gartner Peer Insights had 26.0%, G2 23.1%, and Capterra 17.8%. The data comes from one snapshot, taken on Dec 1, 2025.
Both findings can hold. SE Ranking, which sells SEO tools, sampled a broad set of commercial keywords. Ours is 21 answers to software-buying prompts. Of the three engines, AI Overviews are the least settled on review sites, so treat it as an open question.
Put Review-Site Effort Where AI Engines Look
Your review-site effort belongs on the pages AI engines cite. That means category placement, integration listings, the review sites' own "best X" articles, and accurate pricing. A complete profile is the baseline underneath all of it. In our audit, Perplexity alone cited 19 category or directory pages on review sites, and no vendor review profiles.
Where we'd put the work:
- Category placement. Check that you're listed in every G2 and Capterra category your buyers would browse. Category and directory pages were the review-site pages Perplexity cited most.
- Integration listings. Record every integration you support, on every directory you're listed on. Integration prompts drew pages filtered by a single integration, such as GetApp's lists of project management tools that work with GitHub.
- The sites' own editorial articles. They made up 12 of Perplexity's 37 review-site citations and both of ChatGPT's, and most were "best X" lists. Treat them like any other list you want to appear in.
- Accurate pricing. ChatGPT retrieved 9 product pricing pages on review sites and cited none of them. Keep your profile's prices in line with your own pricing page, which is what ChatGPT tends to cite. Our guide to getting your SaaS cited in ChatGPT and Perplexity covers the owned-page side.
- A complete profile, as the baseline. Fill in the fields and keep them current. Your categories, integrations, and pricing on each site all live there. Treat it as hygiene: in our audit, it was never the page that got cited.
Weight the work by engine. Perplexity is where review sites earned citations, so start there. ChatGPT cited no review profile in 25 answers and only two review-site pages of any kind, so a review drive probably won't move it. It leaned on vendors' own pages, and AI Overviews on lists, Reddit, and YouTube. Our breakdown of how AI engines decide which SaaS vendors to recommend covers those layers.
Do More Reviews Get You Recommended?
Probably not by much. The engines that cited review sites cited their category pages and lists, not the profile where your reviews sit. Our audit has two limits here. It's a single snapshot, and it doesn't record which vendors have profiles or how many reviews they have, so it can't test volume directly.
It helps to separate inclusion from rank. A review profile plausibly helps get a product considered, since the listing is what places it on a category page. If you have no profile, or one filed under the wrong categories, you may be missing from the category pages Perplexity cites. That's worth fixing first.
Rank is a different problem. Once every tool on a shortlist has a profile with plenty of reviews, the count stops separating them. So what would make an engine name one of five well-reviewed CRMs first? The review count alone probably wouldn't. On our prompts, what the engines cited was mostly lists and vendors' own pages. That's why we'd treat a complete profile as hygiene and volume as a weak lever.
Reviews may count for more inside the review sites themselves. A G2 category page or "best X" article is the site's own ordering of products, and where you land there is the site's call. Each site sets its own order, and we won't guess at the formula.
Writing for G2, Kevin Indig analyzed 30,000 AI citations across 500 G2 software categories, using Profound data from ChatGPT, Perplexity, and Claude. "Categories with 10% more reviews have 2% more citations," the analysis found, and "reviews explain less than 2% of the variance." G2 published this analysis, so the finding runs against its own commercial interest. It's also category-level and correlational, so it can't show what more reviews would do for one product.
Frequently Asked Questions
Does ChatGPT use G2 reviews to recommend software?
Very little, on our evidence. ChatGPT cited a review site in 2 of our 25 software-buying answers, and both were learn.g2.com "best X" articles, not product profiles. It retrieved 9 product pricing pages on review sites and cited none of them. According to DerivateX's study, G2 and Capterra received zero ChatGPT citations across 40 software categories.
Does Capterra show up in AI answers?
Mostly on Perplexity. In our audit of 25 software-buying prompts, Perplexity cited Capterra 8 times, second only to G2's 19. None of those 8 was a product's review profile, in line with every other review-site citation we logged. ChatGPT cited no Capterra page, and neither did Google AI Overviews.
Do more G2 reviews improve AI visibility?
Not by much, on the evidence available. According to Kevin Indig's analysis for G2 of 30,000 AI citations, categories with 10% more reviews have 2% more citations. Reviews explain less than 2% of the variance. Our audit can't test volume, but it found engines citing review sites' lists rather than profiles.
Which review-site pages do AI engines cite?
AI engines cite category pages, the sites' own "best X" articles, and comparison pages. Across 41 review-site citations in our audit, none was a vendor's own review profile. Perplexity's 37 included 19 category or directory pages, 12 editorial articles, and 5 comparison or alternatives pages.
Do Google AI Overviews cite review sites?
Rarely, in our audit. Only 2 of 21 answers cited a review site, one a G2 category page and the other a Gartner Peer Insights alternatives page. According to SE Ranking's study of commercial keywords, 34.5% of AI Overview responses cited at least one review platform. Our sample is small, so we treat AI Overviews as the open question.
Conclusion
On the evidence we have, G2 and Capterra reviews don't decide whether AI recommends your SaaS. Review sites earn citations mainly on Perplexity, through their category pages and editorial lists. They barely register in ChatGPT's citations, and Kevin Indig's analysis for G2 ties review volume to little change in citations. Keep a complete profile, get your categories and integrations right, and aim for the sites' own "best X" articles.
Agencies such as Kubnal Bridge specialize in mapping which review-site pages each engine cites for your category. If you're unsure where your product stands, that map is a sensible place to start.
About the Author
Kubnal Bridge Editorial Team. The Kubnal Bridge editorial team researches and writes about generative engine optimization and AI search visibility across ChatGPT, Perplexity, Google AI Overviews, Gemini, and Claude.
Sources
- DerivateX (Apoorv Sharma), "The Authority Inversion: How ChatGPT Really Decides Which Software to Recommend," 2026-06-18, https://derivatex.agency/report/chatgpt-software-recommendation-sources/
- SE Ranking (Yulia Deda), "Despite 90% Traffic Loss, Review Platforms Top AI Overview Citations," 2026-01-29, https://seranking.com/blog/review-platforms-in-ai-overviews/
- G2 Learn (Kevin Indig), "Do More G2 Reviews Mean More AI Visibility? Insights from 30k Citations," 2025-10-23, https://learn.g2.com/do-more-g2-reviews-mean-more-ai-visibility
- Kubnal Bridge, SaaS citation audit: 25 software-buying prompts across ChatGPT, Perplexity, and Google AI Overviews (first-hand data)