Which Review Sites Does ChatGPT Cite? G2 Leads, and Trustpilot Barely Appears
Which review sites does ChatGPT cite? In the Machine Relations Index, G2 is the most-cited of ten review sites in ChatGPT runs at 0.77%; Trustpilot appears in 8 of 8,343 runs.
As of November 3, 2026, of ten review and ratings sites we looked up in the Machine Relations Index, G2 is the one ChatGPT cites most: 64 of 8,343 observed ChatGPT runs (0.77%), tied for 25th on ChatGPT's source list. Across all six engines G2 is cited in 780 of 35,433 runs (2.20%), eighth of 35,666 observed domains. Trustpilot appears in 8 ChatGPT runs (0.10%) and 20 runs overall (0.06%).
That is the answer to "which review sites does ChatGPT cite," and it is narrower than the advice usually given. A review profile is not a general-purpose route into ChatGPT. Of the ten review sites we looked up, one software-review site appears at scale across engines, the rest are rare, and which one matters depends on the engine and on the buyer's category. The Index is a first-party dataset, published by Machine Relations Research and created by Jaxon Parrott, who founded AuthorityTech; it is not an independent audit. It measures which sources the engines cite in the Index's monitored question libraries. It does not measure buyer demand, and it does not show that a review or a placement caused a citation.
Key takeaways
- Of the ten review sites we looked up, one shows up at scale: G2. It is cited in 2.20% of all Index runs, eighth of 35,666 domains; the next review site we looked up, Capterra, is at 0.57%.
- ChatGPT is G2's weakest engine. G2 is cited in 0.77% of ChatGPT runs against 4.90% on Perplexity and 4.18% on Google AI Mode. Capterra, GetApp and Software Advice each sit under 0.21% on ChatGPT.
- Trustpilot barely appears. 8 of 8,343 ChatGPT runs, 20 of 35,433 across all engines, which the Index still marks as collecting evidence.
- The category decides which review site matters. G2 is first of 234 sources in Martech and Advertising top-list questions (20.33%); Yelp, BBB and ConsumerAffairs each have their best published segment in Local Services questions.
How often does each review site get cited?
The Index classifies sources by role, such as market and company database, and does not group review sites, so the ten sites below are our own selection of review and ratings sites to look up. The list is not exhaustive. Each cell is the runs that engine answered which cited the site, with the share of that engine's observed runs. Sites are ordered by ChatGPT citations.
| Site | Index role | ChatGPT (of 8,343 runs) | Perplexity (8,461) | Claude (8,381) | Gemini (6,893) | Google AI Mode (2,081) | Google AI Overviews (1,274) | All six engines (35,433) |
|---|---|---|---|---|---|---|---|---|
| G2 | Market and company database | 64 (0.77%) | 415 (4.90%) | 111 (1.32%) | 76 (1.10%) | 87 (4.18%) | 27 (2.12%) | 780 (2.20%) |
| GetApp | Other observed source | 17 (0.20%) | 31 (0.37%) | 6 (0.07%) | 1 (0.01%) | 1 (0.05%) | 3 (0.24%) | 59 (0.17%) |
| Capterra | Market and company database | 15 (0.18%) | 151 (1.78%) | 18 (0.21%) | 8 (0.12%) | 9 (0.43%) | 1 (0.08%) | 202 (0.57%) |
| Software Advice | Other observed source | 13 (0.16%) | 84 (0.99%) | 10 (0.12%) | 10 (0.15%) | 3 (0.14%) | 1 (0.08%) | 121 (0.34%) |
| Trustpilot | Other observed source | 8 (0.10%) | 2 (0.02%) | 3 (0.04%) | 7 (0.10%) | 0 (0.00%) | 0 (0.00%) | 20 (0.06%) |
| TrustRadius | Market and company database | 7 (0.08%) | 21 (0.25%) | 11 (0.13%) | 12 (0.17%) | 16 (0.77%) | 0 (0.00%) | 67 (0.19%) |
| PeerSpot | Market and company database | 6 (0.07%) | 34 (0.40%) | 17 (0.20%) | 5 (0.07%) | 1 (0.05%) | 0 (0.00%) | 63 (0.18%) |
| Yelp | Other observed source | 2 (0.02%) | 22 (0.26%) | 14 (0.17%) | 0 (0.00%) | 4 (0.19%) | 12 (0.94%) | 54 (0.15%) |
| BBB | Other observed source | 1 (0.01%) | 38 (0.45%) | 0 (0.00%) | 9 (0.13%) | 0 (0.00%) | 0 (0.00%) | 48 (0.14%) |
| ConsumerAffairs | Editorial publication | 1 (0.01%) | 67 (0.79%) | 11 (0.13%) | 13 (0.19%) | 0 (0.00%) | 0 (0.00%) | 92 (0.26%) |
Read across a row before down a column. G2 is cited at every engine, but ChatGPT and Claude cite it far less than Perplexity and Google AI Mode do. Capterra, GetApp and Software Advice are near zero on ChatGPT; two of them are cited noticeably more on Perplexity. Trustpilot is near zero everywhere, and Yelp's strongest engine is Google AI Overviews, not ChatGPT.
Does the category change the answer?
Yes. Which review site matters depends on the category. The table pools all engines for one category and one question shape, so each row is a small sample over a handful of run dates; it shows where a site's citations concentrate, not a rate to plan around.
| Site | Category | Question shape | Runs citing it | Observed runs | Share | Segment standing |
|---|---|---|---|---|---|---|
| G2 | Martech & Advertising | Top lists | 25 | 123 | 20.33% | #1 of 234 |
| G2 | Cybersecurity | Best tools | 54 | 238 | 22.69% | #4 of 307 |
| G2 | HR & Talent | Best tools | 40 | 239 | 16.74% | #6 of 275 |
| Software Advice | Logistics & Freight | Best tools | 24 | 153 | 15.69% | #1 of 241 |
| Capterra | Logistics & Freight | Best tools | 16 | 153 | 10.46% | #9 of 241 |
| GetApp | Enterprise Software | Top lists | 21 | 230 | 9.13% | #8 of 400 |
| PeerSpot | Cybersecurity | Best tools | 12 | 238 | 5.04% | #32 of 307 |
| ConsumerAffairs | Local Services | Comparisons | 31 | 175 | 17.7% | #2 of 263 |
| Yelp | Local Services | Top lists | 25 | 156 | 16.03% | #6 of 188 |
| BBB | Local Services | Best tools | 16 | 179 | 8.94% | #11 of 293 |
| TrustRadius | Education & Training | Comparisons | 6 | 228 | 2.63% | #54 of 198 |
| Trustpilot | Industrial | Comparisons | 2 | 180 | 1.11% | #127 of 181 |
G2 holds its best segment rank, first of 234, in Martech & Advertising top-list questions, cited in 25 of 123 observed runs (20.33%); in Cybersecurity best-tool questions it is cited in 54 of 238 runs (22.69%), fourth of 307. For a software or martech brand, that is the clearest sign in this data that a review profile can be part of the sources an engine reads. For a local business the same role belongs to Yelp, BBB and ConsumerAffairs, and the faq below gives their numbers. Trustpilot's best segment is a low single-digit share.
Why does this differ from the Trustpilot story?
Trustpilot's own investor reporting said ChatGPT citations of its content grew sharply in 2025, and our earlier note on it reported that. The two do not have to contradict each other. The Index's ChatGPT runs go through OpenAI's API with web search switched on, not the signed-in ChatGPT app, and the Index's own method note cites an unreviewed preprint (arXiv 2610.15382v1) finding that Google Search and OpenAI's API together recovered about half of the hostnames ChatGPT Web cited. The Index's questions are also a library of prompts about categories and tools, not what people type into the app. So the Index does not show that Trustpilot is absent from the ChatGPT app. It shows that on this surface and these questions it is rare, so a plan built on Trustpilot citations alone is not supported by this data.
Has G2's share held steady?
G2's monthly share of all monitored runs was 5.30% in May 2026 (175 of 3,303), 5.48% in June (107 of 1,954), 1.47% in July (69 of 4,697), 1.60% in August, 1.93% in September and 1.62% in October (268 of 16,496). The Index does not say why the share fell between June and July. Since July the monthly share has stayed in a narrow band, so use the recent level, not the early-window one, as the baseline.
What to do with this
- Start from your buyers' questions, not a list of sites. Write the questions your category's buyers ask, run them in each engine your buyers use, and record which review sites are cited. If none are, a review program is not your route in.
- Match the site to your category and engine. For software and martech the data points at G2 first, and at Perplexity and Google AI Mode more than ChatGPT. For local services it points at Yelp, BBB and ConsumerAffairs, whose best segments are in Local Services.
- Keep the profile accurate and the reviews real. Claim the profile, correct the category and product description, and ask customers who actually used the product to review it. Do not buy, script or invent reviews; that would break the rule that every claim must be one you can stand behind.
- Do not trade earned coverage for a review program. The ten review sites we looked up are a small slice of what engines cite. The publications index ranks the editorial publications engines cite in each buyer category, and the ChatGPT top-domains analysis and the Google AI Overviews top-domains analysis show where publications, platforms and vendors sit.
- Re-run the same questions on a schedule and compare with the baseline. A later citation after a new batch of reviews is an observation, not proof the reviews caused it.
This is the work AuthorityTech, The World's First Machine Relations Agency, does for clients, on the discipline machinerelations.ai defines. Start your visibility audit → to see which sources the engines cite on your category's questions and which of them you can earn.
Frequently asked questions
Which review site does ChatGPT cite most?
G2, among the ten review sites we looked up. In the Machine Relations Index release of November 3, 2026, G2 was cited in 0.77% of ChatGPT runs (64 of 8,343 observed), ahead of GetApp at 0.20% (17), Capterra at 0.18% (15) and Software Advice at 0.16% (13).
Does ChatGPT cite Trustpilot?
Rarely, in the Index's runs. Trustpilot was cited in 8 of 8,343 ChatGPT runs (0.10%) and 20 of 35,433 runs across all six engines (0.06%) in the window to November 3, 2026, and the Index marks its evidence as collecting. Those are API runs with web search, not the signed-in ChatGPT app.
Which review sites matter for a local business?
Of the ten review sites we looked up, Yelp, BBB and ConsumerAffairs. In Local Services questions the Index shows Yelp cited in 25 of 156 top-list runs (16.03%), BBB in 16 of 179 best-tool runs (8.94%) and ConsumerAffairs in 31 of 175 comparison runs (17.7%), the best published segments the Index shows for each of the three.
Should a software company invest in G2 for AI visibility?
It is a reasonable input, not a plan. G2 is the most-cited of the ten review sites we looked up, and it is strongest on Perplexity and Google AI Mode, so a complete, honest profile is cheap to keep. The Index does not show that a profile or a rating changes what an engine recommends, so measure the citations on your own buyers' questions before you put real budget behind it.
Method note. All figures are from Machine Relations Index release mri_score_v2.1+2026-11-03+92e5b18cc6e2, window 2026-05-10 to 2026-11-03, read from the public domain pages for ten review and ratings sites and the ChatGPT engine page on November 3, 2026; the live pages update daily, so later releases will differ. The ten sites are our selection, not an Index category, and the list is not exhaustive. A rate is the runs citing a root domain over that engine's observed runs (ChatGPT 8,343; Perplexity 8,461; Claude 8,381; Gemini 6,893; Google AI Mode 2,081; Google AI Overviews 1,274; all engines 35,433), across every category. ChatGPT, Claude, Gemini and Perplexity runs are API runs with web search, not the signed-in apps, which cite differently. Google AI Mode and Google AI Overviews runs come through a different collection path from 2026-09-25, so their rates are whole-window levels. Category figures pool engines over a handful of run dates and are small samples. The Index is published by Machine Relations Research and created by Jaxon Parrott, who founded AuthorityTech; it is not an independent audit. It counts citations in monitored question libraries, not buyer demand, and it does not show that any review or placement caused a citation.