Pave was the product ChatGPT and Gemini named most often when we asked them to recommend compensation management software, appearing in 5 of the 6 buyer questions we ran (83%). No other product came close on reach. On 2026-08-06 we put six real buyer questions to the two assistants, gave each question one live web search, and recorded every product named and every source domain behind the answers.
Two products tied for second: Beqom and Comprehensive were each named in 4 of the 6 answers (67%). Three more clustered at 3 of 6 (50%): Payfactors, Payscale and Ravio. Everything else, 14 more distinct labels, was named in 2 answers or fewer, and 6 of those labels showed up in just a single question.
The shortlist changed with the wording of the question
This is the section worth dwelling on. Buyers do not ask one question, they ask several, and our six barely agreed with each other: no single product was named in all 6, and even Pave, the leader, missed one question (named in 5 of 6).
The plain query, "best compensation management software," returned just five names: PeopleFluent Compensation Management, BambooHR, Paylocity, Compport and Comprehensive. Pave, the product named in 5 of the other answers, did not appear in the plain "best" answer at all.
The "best free" and "most affordable" questions behaved almost like one question. Of the 8 products each returned, 6 were shared: CaptivateIQ, Beqom, Pave, Payscale, Lattice and Deel appeared in both. "Best free" added Ravio and Salary.com's CompAnalyst label; "most affordable" added Payfactors and the plain Salary.com label. Price framing pulled a steady cluster of 6 that the bare "best" framing never produced.
Comprehensive, named in 4 of 6 answers overall, was absent from both price questions. It showed up under "best," "small teams," "what should I use" and "recommendations," but under neither "free" nor "affordable," a reminder that a strong overall score can still miss whole slices of buyer intent.
Six products were named by only one question each, which shows how thin some placements are. 15Five and ADP appeared only under "best for small teams." Paylocity and PeopleFluent Compensation Management appeared only under the bare "best." Payscale MarketPay appeared only under "what compensation management software should I use," and CompAnalyst appeared only under "recommendations." One reword would have dropped each of them from the results entirely.
Where the answers came from
Three domains fed all 6 answers: bamboohr.com, comprehensive.io and gartner.com. Gartner in every answer is the analyst-report pattern you would expect, but the other two are vendors' own websites, not neutral directories, and that turns out to be the whole story.
The clearest signal in the measurement is that a vendor's own domain fed the answers and that same vendor got named. comprehensive.io was cited in all 6 answers and Comprehensive was named in 4 of them. bamboohr.com was cited in all 6 and BambooHR was named in 2. The pairing repeats for Ravio (ravio.com in 2 answers, named in 3), CaptivateIQ (captivateiq.com in 2, named in 2), plus PeopleFluent, ADP and Paylocity. Seven of the named products had their own site feeding the assistants.
Owning a source is not a guarantee by itself: opencomp.com was cited in 1 answer but OpenComp was never named. Even so, being one of the sources looks close to a precondition, because almost every named product had a page in the pool.
Review directories showed up unevenly. G2 was a source in 5 of 6 answers as g2.com, and its learn.g2.com subdomain fed the sixth (the "best free" question), so the G2 brand touched all 6 answers. Capterra did not appear once. Reddit was the only community source, cited in 5 of 6 answers, absent only from "best free." Quora and other forums did not appear at all. The long tail of single-citation sources was hrdatahub.com, learn.g2.com, opencomp.com and paylocity.com, one answer each.
What this means if you sell compensation software
The source list, not your market share, decided who got named. The assistants named products they could find in Gartner, G2, Reddit and a short list of vendor sites. A larger competitor that was not in those specific pages did not get named more often; presence in the sources is what moved the outcome.
Comprehensive is the example to study. It is not the biggest name in the category, yet its own site was cited in all 6 answers and it was named in 4, tying Beqom for second overall. Publishing your own structured comparison content, and getting into the Gartner and G2 pages these assistants lean on, is the concrete play.
The practical audit is short: check whether you appear on gartner.com and g2.com (each feeding 5 to 6 answers), on reddit.com (5 answers), and on your own domain in a form an assistant will quote. Capterra fed 0 answers in this run, so time spent there would not have changed a single recommendation.
The full list, counted
| Product | Named in | Share |
|---|---|---|
| Pave | 5 of 6 | 83% |
| Beqom | 4 of 6 | 67% |
| Comprehensive | 4 of 6 | 67% |
| Payfactors | 3 of 6 | 50% |
| Payscale | 3 of 6 | 50% |
| Ravio | 3 of 6 | 50% |
| BambooHR | 2 of 6 | 33% |
| CaptivateIQ | 2 of 6 | 33% |
| Compport | 2 of 6 | 33% |
| Deel | 2 of 6 | 33% |
| HRSoft | 2 of 6 | 33% |
| Lattice | 2 of 6 | 33% |
| Salary.com | 2 of 6 | 33% |
| Salary.com (CompAnalyst) | 2 of 6 | 33% |
Two brands were split across more than one label, so we kept them as separate rows rather than summing counts the data does not show. Salary.com's tool appeared under three labels, Salary.com (2), Salary.com (CompAnalyst) (2) and CompAnalyst (1), which is one vendor's product line, not three. Payscale appeared as both Payscale (3) and Payscale MarketPay (1). Read each row at its measured count, not as separate products.
Where Glotier fits, and where it does not
Glotier does not sell compensation management software, so we were named in 0 of these 6 answers, which is the correct result. We would not want an assistant inventing us into a category we are not in.
This measurement is exactly what we run for the category you do sell in, though. We put the same six-question fan-out to ChatGPT and Gemini, count who gets named out of 6, and map the source pages behind every answer, so you can see whether you are in the pool that gets recommended. The check is free, there is no account, and it takes about a minute.