GEO study: which HR software gets mentioned in AI answers in the US?
We put 90 questions about HR software in the United States to five AI engines and probed 21 company websites. The engines rank the brands in nearly the same order and differ on who is named for any one question. A generative engine optimisation (GEO) study.
AI, TELL ME! HR Software Visibility Study, United States. By the AI, TELL ME! team. Published November 2026.
Short answer. Among the HR software brands the dashboard tracked, BambooHR, Rippling, Workday, Gusto and Lattice hold 52.2% of the 4,549 mentions in 450 answers from five AI engines, and the engines rank the brands in nearly the same order. For any one question the five answers agree less: 13.5% of brand-question pairs were named by all five engines and 42.6% by one engine only. Who appears depends on the segment: Lattice is named in 125 of 150 Employee Performance & Development answers and in 3 of the other 300. Companies supply 72.6% of the cited items, and a named brand’s own site is cited in the same answer in 13.8% of brand-answer pairs, counting answers without source links. Across the 16 company websites that could be read (of 21 probed), no measured site property showed a detectable link to being named, and the sample cannot rule one out.

Ask five AI engines the same question and the lists differ. As an illustration, we put one of the 90 questions to ChatGPT, Gemini, Google AI Mode, Google AI Overview and Claude: “What HR platforms bring employee records, payroll, and benefits administration together?” ChatGPT named 7 HR platforms, Gemini 11, Google AI Mode 4, Google AI Overview 5 and Claude 13. Across the five answers 16 different brands appeared, three of them (BambooHR, Gusto and Rippling) in every answer and 5 in one answer each (Figure 1).

Figure 1. Brands named by ChatGPT, Gemini, Google AI Mode, Google AI Overview and Claude in answer to “What HR platforms bring employee records, payroll, and benefits administration together?” Brands outside the 15 named most often in the whole study read “other brand”.
BambooHR, Gusto and Rippling, the three brands every engine named in Figure 1, are also among the five most-mentioned brands of the whole study. The ranking across all 450 answers holds steady, and a single answer varies.
Five engines, three segments and one set of rules for counting
AI, TELL ME! is a full-cycle AEO (answer engine optimisation) and GEO (generative engine optimisation) agency. The monitoring half of this study ran on our own monitoring platform, which records what each AI engine says and cites, and the website half ran on our AI Access Report, which records what a probe using the engines’ crawler names can reach and read on a website.
The monitoring ran in English with a United States geo target on 4 November 2026. It covers three segments (HR Management & Automation, Recruiting & Onboarding and Employee Performance & Development) with 30 questions each, put once to each of five engines (ChatGPT, Gemini, Google AI Mode, Google AI Overview and Claude), which makes 450 answers, 150 per segment and 90 per engine. In the first segment, for example, a question reads “What HR systems should we look at for a company in the US?”
The dashboard behind the monitoring tracks a fixed list of brands for the category, so every count of brands below is a count among the brands the dashboard tracked. The website half probed 21 company websites on 4 and 5 November 2026: 14 belonging to brands that appear in the answers and 7 to brands that never do.
The counts in this study use five terms:
A mention is one count of a brand in an answer, and one answer can carry several mentions of the same brand.
Answers naming a brand counts the answers in which the platform’s brand extraction recorded the brand at least once, out of 450.
A brand-answer pair is one brand named in one answer.
A brand-question pair is one brand named in the answers to one question, counted once however many engines name it.
A cited item is one cited domain in one answer, with repeats inside an answer counted.
Four questions about AI visibility, and what the data says
Table 1. The four questions this study asks about AI visibility of HR software brands, each with a short answer.
Question | Short answer |
|---|---|
Do the five engines agree on which HR software to name? | On the brand table, yes: rank correlations between engines run from 0.82 to 0.93. On a single question, rarely: 13.5% of brand-question pairs were named by all five engines. |
Is there one leaderboard for HR software? | No. Brand visibility differs by segment, and Lattice is named in more Employee Performance & Development answers than any other brand. |
Do the answers cite the named brand’s own site? | Seldom: in 13.8% of brand-answer pairs, counting answers without source links. |
Does a better-built website get a brand named more often? | Not in this sample. No measured site property showed a detectable link to being named. |
Five brands hold just over half of all mentions
Among the brands the dashboard tracked, BambooHR, Rippling, Workday, Gusto and Lattice hold 52.2% of the 4,549 mentions, and BambooHR alone holds 15.9%. The next five brands hold 20.6% and the other 43 share 27.2%. The same five lead by answers naming a brand (Table 2). The first four keep their places on every counting basis we tested, and the order of the fifth to the seventh place (Lattice, HiBob and ADP) changes with the counting basis.
Table 2. The ten most-mentioned HR software brands among those the dashboard tracked, with mentions, share of the 4,549 mentions and answers naming the brand out of 450.
Brand | Mentions | Share of mentions | Answers naming the brand (of 450) |
|---|---|---|---|
BambooHR | 724 | 15.9% | 374 |
Rippling | 652 | 14.3% | 308 |
Workday | 399 | 8.8% | 208 |
Gusto | 312 | 6.9% | 165 |
Lattice | 287 | 6.3% | 128 |
HiBob | 231 | 5.1% | 121 |
ADP | 216 | 4.7% | 120 |
15Five | 178 | 3.9% | 101 |
Leapsome | 175 | 3.8% | 86 |
Paylocity | 136 | 3.0% | 94 |
The leader depends on the measure and the engine
BambooHR and Rippling hold the top two places by mentions on every engine, and the order between them changes. BambooHR leads on ChatGPT, Google AI Overview and Claude, and Rippling leads on Gemini and Google AI Mode. By answers naming a brand BambooHR leads on every engine, from 70 of 90 answers on Google AI Overview to 82 on Claude.
Twenty-four of the 53 brands appear in nine answers or fewer
Behind the leaders sits a long tail. Among the 53 brands the dashboard tracked and the engines named, 24 are named in nine answers or fewer, 12 of them in exactly one answer, and together the 24 hold 2.1% of the mentions.
The engines rank the brands alike and answer one question differently
The five engines rank the 53 brands in nearly the same order (rank correlations of 0.82 to 0.93), and for a single question 13.5% of brand-question pairs were named by all five engines.
The rank order is nearly the same on every engine
Rank correlations of mentions between engines run from 0.82 (Google AI Mode with Claude) to 0.93 (ChatGPT with Gemini) across the 53 brands, all ten pairs at p < 0.0001. Of the 53 brands, 25 are named on all five engines and hold 95.7% of the mentions, and 13 are named on one engine only, with 20 mentions between them.
About one pair in seven is named by all five engines
Across the 90 questions the engines produced 1,065 brand-question pairs among the tracked brands. About one pair in seven (13.5%) was named by all five engines and more than two in five (42.6%) by one engine only (Figure 2). In the median question the five answers name 12 different tracked brands and only 2 of them appear in all five, and in 18 of the 90 questions no tracked brand is named by all five engines.

Figure 2. Share of the 1,065 brand-question pairs by the number of engines that named the brand, for all pairs, for the five leading brands by mentions (BambooHR, Rippling, Workday, Gusto and Lattice) and for the other 48 brands.
The agreement sits with the leaders. The five leading brands by mentions (BambooHR, Rippling, Workday, Gusto and Lattice) account for 351 pairs, 33.9% of them named by all five engines, and they hold 119 of the 144 all-five pairs (82.6%). The other 48 brands account for 714 pairs, 3.5% of them named by all five and 54.9% by one engine only.
Claude accounts for 188 of the 454 one-engine pairs, and without Claude 45.0% of the remaining 877 pairs are still named by one of the four engines only. Because each engine answered each question once, the study makes no claim about which engine prefers which brand.
Claude names more brands per answer, and five of the ten most-named brands vary by engine
Claude’s answers name 8 brands at the median against 5 on each of the other four engines (Table 3; Mann-Whitney p < 0.0001).
Table 3. Engine profile: mentions, brands named, median brands per answer and share held by the engine’s five most-mentioned brands, for each of the five engines and for all engines together.
Engine | Mentions | Tracked brands named | Median brands per answer | Share held by the engine’s five most-mentioned brands |
|---|---|---|---|---|
ChatGPT | 1,228 | 30 | 5 | 50.7% |
Gemini | 985 | 38 | 5 | 59.6% |
Google AI Mode | 629 | 32 | 5 | 56.6% |
Google AI Overview | 576 | 41 | 5 | 55.4% |
Claude | 1,131 | 42 | 8 | 47.8% |
All engines | 4,549 | 53 | 5 | 52.2% |
The engine matters for some brands. Five of the ten most-named brands differ across the engines past a Bonferroni threshold of 0.005: HiBob, which is named in 4 to 34 of the 90 answers on an engine, and ADP, Paylocity, Gusto and Workday. BambooHR shows no detectable difference (p = 0.18) and Rippling sits just above the threshold (p = 0.0057).
Brand visibility differs by segment
The brands change with the part of HR a question is about. Nine of the ten most-named brands differ across the three segments past a Bonferroni threshold of 0.005, and Lattice differs most: it is named in 125 of 150 Employee Performance & Development answers and in 3 of the other 300 (Figure 3).

Figure 3. Answers naming each of the ten most-named brands in the three segments (HR Management & Automation, Recruiting & Onboarding and Employee Performance & Development), out of 150 per segment, with the share of brand-question pairs named by all five engines in the bottom row.
Specialists flip between segments and Workday stays level
The specialist 15Five, like Lattice, is named in 100 Employee Performance & Development answers and in 1 of the other 300. Gusto falls from 103 answers in HR Management & Automation to 44 in Recruiting & Onboarding and 18 in Employee Performance & Development.
BambooHR is named in more answers than any other tracked brand in two of the three segments and in 71% to 93% of the answers in every segment. Workday, named in 64, 74 and 70 answers, is the only one of the ten with no detectable difference (p = 0.51). A single overall visibility figure for a brand would hide these splits.
In Recruiting & Onboarding, the tracked brands named in the most answers are BambooHR (128 of 150) and Rippling (107), two general HR platforms, followed by Workable (76) and Workday (74).
The engines agree most on HR management questions and least on performance questions
Among the tracked brands, the share of brand-question pairs named by all five engines is 18.8% in HR Management & Automation (68 of 361), 13.0% in Recruiting & Onboarding (39 of 301) and 9.2% in Employee Performance & Development (37 of 403). The difference holds in a chi-square test (p = 0.0005) and in a permutation of segment labels over the 90 questions (p < 0.001).
Companies supply almost three quarters of the citations
Most of what the engines cite comes from companies. Companies that sell or serve in the HR segments supply 72.6% of the 3,233 cited items, which sit on 414 domains in 382 of the 450 answers; the other answers carried no source links.
Three kinds of company make up that share:
The tracked brands’ own sites supply 25.2% of all items, or 22.4% without peoplegoal.com, which alone supplies 92 items.
Vendors the dashboard does not track supply 38.9%.
Agencies, consultancies, recruiters and training providers supply 8.4%.
Table 4 and Figure 4 give the other source types.
Table 4. Cited items in the 450 answers by source type, classified by the business of each domain’s owner.
Source type | Domains | Cited items | Share of items | Answers citing at least one (of 382) |
|---|---|---|---|---|
Own sites of the tracked brands | 43 | 816 | 25.2% | 231 |
Vendors the dashboard does not track | 223 | 1,259 | 38.9% | 293 |
Agencies, consultancies, recruiters and training providers | 46 | 271 | 8.4% | 106 |
Review, comparison and directory sites | 45 | 392 | 12.1% | 153 |
Media and publishers | 35 | 264 | 8.2% | 113 |
Community, video and social platforms | 10 | 194 | 6.0% | 76 |
Institutions, infrastructure and other | 12 | 37 | 1.1% | 17 |
All cited items | 414 | 3,233 | 100% | 382 |

Figure 4. Share of cited items by source type, all engines and each engine, 3,233 items on 414 domains. Companies are the first three types added together.
The source mix varies by engine
Each engine leans on a different mix. Community, video and social platforms are 19.7% of Google AI Overview’s items (162 of 823) and none of ChatGPT’s 372, and YouTube alone supplies 124 items in Google AI Overview answers. Review and directory sites are 29.0% of ChatGPT’s items against 5.6% on both Google AI Mode and Google AI Overview. Companies supply the majority on every engine, from 62.9% of ChatGPT’s items to 80.8% of Google AI Mode’s, and because items inside one answer are not independent these percentages carry no test.
Citations are spread thin
Half of the 414 cited domains appear in one answer only (211, or 51.0%). By cited items, the leading domain differs on every engine: technologyadvice.com on ChatGPT, g2.com on Gemini, tabulera.com on Google AI Mode, youtube.com on Google AI Overview and peoplemanagingpeople.com on Claude.
A brand’s own site is cited in the same answer in about one pair in seven
When an engine names a brand, its own site is seldom cited in the same answer: in 344 of 2,501 brand-answer pairs (13.8%), a base that includes the 68 answers without source links. Among the 382 answers with source links the share is 344 of 2,136 (16.1%). Among the ten most-named brands the share runs from 21.9% for Lattice to 8.2% for Workday (Figure 5).

Figure 5. Share of the answers naming each of the ten most-named brands in which a domain owned by that brand is also cited, with the share for all 53 brands as a line.
Brands named more often also have their own sites cited more often (rank correlation of mentions with own-site cited items 0.76, p < 0.0001, n = 53). The relation is partly mechanical, because a brand named in more answers has more answers in which its own site can be cited.
Most websites served the AI citation crawlers, and structure was the lowest of the four page-based scores
Of the 21 company websites, 16 served the four crawler identities behind ChatGPT and Claude (OAI-SearchBot, ChatGPT-User, Claude-SearchBot and Claude-User). Fifteen served all 8 citation identities and 8 served all 14 identities.
The page-based scores (text without JavaScript, content extraction, passage quality and structure) need a page read, and among them structure is lowest, with a median of 67 and a range of 10 to 92. Site results appear as counts and ranges, and none is tied to a brand, a share or a rank.

Figure 6. Six scores for the company websites that could be read, one mark per site, with the lowest to highest score as a band and the median as a line. Sites are anonymous, and the five sites with no page are not plotted.
Five sites could not be measured
The bot protection of five of the 21 company websites stopped an automated request, so the probe has no verdict: all 14 crawler identities received an error reply and no page was read. The sites carry no page-based score and appear as not measured, never as zero. One of the five is cited as a source, in one answer.
Crawler replies are reported as counts
Of the 21 company websites, 8 refused at least one training crawler by name, and 7 of those 8 still served all 8 citation identities. The robots.txt file, readable at 19 sites, disallows none of the 16 names tested. A probe shows what a test client received, and the 14 identities include no Googlebot, so none of these results speaks to Gemini, Google AI Mode or Google AI Overview.
Table 5. Crawler reply patterns of the 21 company websites to the AI Access Report probe, as counts of sites.
Reply pattern | Sites |
|---|---|
Served every identity | 8 |
Refused one training crawler by name (Bytespider) | 5 |
Refused three training crawlers by name | 1 |
Refused five training crawlers by name, although robots.txt allows them | 1 |
Refused nine crawlers, among them three citation crawlers | 1 |
Refused every identity and returned no page | 5 |
Access and text sit near the ceiling, and structure is the lowest of the four page-based scores
Crawler access and text without JavaScript sit near the ceiling, with medians of 99 and 100. Passage quality has a median of 96, content extraction 87 and structure 67, and lab speed has the lowest median of the six scores, 59. On the three pages tested, one site held no text before scripts ran and has the lowest text-without-JavaScript and structure scores.
The page signals of the 16 sites with pages read, on 20 to 21 sampled pages each:
JSON-LD sits on every sampled page at 3 sites, on some pages at 10 and on none at 3.
A machine-readable date sits on some pages at 10 sites and on none at 6, and heading levels are skipped on at least one sampled page at 15 of the 16 sites.
In this sample, no site property shows a detectable link to being named
Seven brands in the site set are named in none of the 450 answers, answer text included. We chose them ourselves, so they are not a sample of the brands the answers never name. Each description comes from the company’s own site:
APS Payroll sells human capital management software covering payroll, HR and attendance.
Ascentis sells HR and workforce software with time tracking, payroll, benefits and hiring tools.
Asure Software sells payroll, HR, benefits and compliance software and services, mainly to small and mid-sized employers.
BerniePortal sells an HRIS with payroll, benefits, onboarding and time tools for small and mid-sized businesses.
ClayHR sells configurable HR software for hiring, onboarding, performance, learning and leave.
Inova Payroll sells payroll services with integrated HR and benefits software.
Patriot Software sells online accounting and payroll software to small businesses.
Their sites went through the same probe as the sites of the brands the answers name, and Figure 6 reports every readable site together.
We tested whether any site property goes with being named. Across 14 site measures and two outcomes (mentions and answers naming a brand) there are 28 tests on the sites of the named brands, and none of the correlations passes a Bonferroni threshold of 0.0018.
Two reach p < 0.05 against 1.4 expected by chance. Both are negative and both are for one measure, the share of sampled pages with unskipped heading levels, set against the two outcomes (rho −0.64 against mentions, p = 0.014; rho −0.60 against answers naming, p = 0.022; n = 14).
The sample shows no detectable link between the measured site properties and being named, and it cannot rule one out.
Implications for an HR software brand
The data leave open whether copying the leaders works
Nothing in this study shows that copying a leading brand’s website would win mentions, and nothing shows that it would fail. Three findings keep the question open:
Being named and being cited differ: Gusto is named in 165 answers and has its own site cited in 15.
Other sources carry most of the citations: the tracked brands’ own sites supply 25.2% of the cited items.
One answer is a poor guide to the overall brand ranking: 13.5% of brand-question pairs were named by all five engines, so a brand missing from one engine’s answer can appear in another’s.
A brand that wants to know where it stands has to measure each segment and each engine and to repeat the measurement on the same questions.
What happens after an audit
For a single brand, AI, TELL ME! runs the work that follows a study of this kind, and monitoring then repeats on the same question set:
AEO Monitoring shows where a brand appears in AI-generated answers and whether it appears at all, with the competitors that appear beside it and the sources cited.
The GEO Report assesses the brand’s website, content and external presence and shows what helps the brand appear in AI recommendations and what holds it back.
The AEO & GEO Roadmap turns the findings into a prioritised 90-day plan for technical, content and external presence work, in order of impact.
Website and content work gives AI systems a clear, accurate picture of what the brand offers, and is delivered with the brand’s own team, fully outsourced or as a mix of both.
PR for AI search builds the brand’s presence in the sources AI systems use to describe it, with a target list of publications and expert voices, published coverage and tracking.
These findings against other published research
Five published analyses bear on these findings, each from a vendor or an agency, and none measures the same thing under the same definitions as this study.
Rand Fishkin of SparkToro and Patrick O’Donnell of Gumshoe (27 January 2026) ran 12 prompts 60 to 100 times each through ChatGPT, Claude and Google’s AI Overview and AI Mode, 2,961 runs in all, and found less than a 1 in 100 chance that ChatGPT or Google’s AI returns the same list of brands in any two responses (Claude repeats a list slightly more often). We asked 90 different questions once each, and the five answers to one question also differ.
DerivateX (29 June 2026), an SEO and GEO agency, compared ChatGPT with Google AI Overviews on 15 queries in cloud media infrastructure, ran each query 5 to 10 times per engine and reported a brand overlap of 32% and a source overlap of 4% on its open-ended queries. Measured as the Jaccard overlap per question averaged over questions, our 90 questions, each asked once, give 37.9% for brands and 2.5% for cited domains.
Trakkr (12 March 2026), an AI visibility monitoring vendor, found that up to eight models name the same top brand in 4.0% of queries. The first brand in the text of our five answers is the same on all five engines in 4 of 87 questions (4.6%), an indicative count with a different definition of the top brand.
Overthink Group (14 July 2026), an agency, analysed 250 B2B software categories on four engines and found YouTube at 7.4% of Google AI Overview citations and Reddit at 1.4% of all citations. In this study YouTube is 15.1% of Google AI Overview items and Reddit 1.0% of all items (33).
Ahrefs (11 May 2026), an SEO software vendor, tested 1,885 pages that added JSON-LD and found a statistically significant 4.6% fall in Google AI Overview citations against matched control pages, which were also already declining, and rises of 2.4% in Google AI Mode and 2.2% in ChatGPT that cannot be told from zero. That is a before and after test on pages that were already cited, and ours compares 16 sites at one point in time.
Frequently asked questions
Do AI assistants recommend the same HR software?
The five engines in this study rank the brands in nearly the same order (rank correlations from 0.82 to 0.93) and name different brands for a single question: 13.5% of the 1,065 brand-question pairs were named by all five engines and 42.6% by one engine only. A listed brand is not necessarily a recommended one, because the pairs do not separate the two.
Which HR software do AI assistants name most often in the US?
Among the 53 brands the dashboard tracked and the engines named, BambooHR is named in the most answers (374 of 450) and has the most mentions (724 of 4,549), followed by Rippling, Workday, Gusto and Lattice. Together the five hold 52.2% of the mentions.
Do AI assistants name different software for HR management, recruiting and performance?
Yes, the brands named differ by segment: Lattice is named in 125 of 150 Employee Performance & Development answers and in 3 of the other 300.
Do AI assistants cite the software company’s own website?
Seldom for the brand being named: a brand’s own site is cited in the same answer in 344 of 2,501 brand-answer pairs (13.8%, counting answers without source links). Companies of all kinds supply 72.6% of the cited items.
Does a better-built website make HR software more visible in AI answers?
No link could be detected in this sample: none of the 28 site-property tests passes the corrected threshold of 0.0018, and the two that reach p < 0.05 are one measure, skipped heading levels, set against two outcomes.
Methodology and limitations
This study covers one category (HR software) in one market (the United States): 450 answers naming 53 brands, 4,549 mentions, 3,233 cited items and 21 company websites, 16 of them read.
Each engine answered each question once, so differences between engines and differences between runs of one engine cannot be told apart. The 90 questions are written in the words of an HR buyer, and another wording could change the ranking.
Mentions divide by 4,549, answers naming a brand by 450 (150 within a segment, 90 within an engine) and source shares by the 3,233 cited items. Brand-question pairs (1,065) and brand-answer pairs (2,501) have their own bases, and 382 of the 450 answers carry source links. The 53 brands are those the dashboard tracked and the engines named; the answers name other products as well, and the study does not claim a complete list of brands for any segment.
Segment and engine comparisons are chi-square tests on the ten most-named brands with a Bonferroni threshold of 0.005. The question-level agreement by segment is a chi-square test checked by a permutation of segment labels (5,000 shuffles). Rank correlations are Spearman coefficients and brands per answer are compared with a Mann-Whitney test. The 28 site tests use a threshold of 0.0018.
Source types are our classification of each cited domain into seven groups by its owner’s business. A cited domain counts as a brand’s own site when its name identifies the brand, and two domains (20 of the 816 own-site items) are matched through a product page or a parent company.
Each of the 21 company websites received one AI Access Report: 14 crawler identities (8 that fetch pages to cite and 6 that fetch pages for training) and 2 robots.txt opt-out tokens, sent to the home page from one address, plus a sample of 20 to 21 sitemap pages. The probes ran on 4 and 5 November 2026 on global company domains, after the answers were collected. Scores are a relative scale for tracking change and are not calibrated against citation. Two further brands named in the answers returned no page to the probe and are not in the set of 21.
Thirty questions per segment is a test scale. Working monitoring runs 50 prompts per segment and monitors every segment on its own, because presence differs between segments (Lattice is named in 125 of 150 Employee Performance & Development answers and in 3 of the other 300).
The figures describe this run: United States, English, three segments, five engines, 4 November 2026 for the monitoring and 4 to 5 November 2026 for the website measurement. The study rates no product.
Running a GEO study for your own category
This study covers HR software in the United States. The same measurement runs for any category, market and language, on the same two halves: what the engines say and cite, and what a probe using their crawler names can reach and read. We bring 17+ years in marketing, PR and SEO to the work.
Book a 30-minute review of your own category at https://calendly.com/natali-khld/welcome.
Request the prompt set, the engine-by-engine breakdown or the AI Access Report for your own website at hey@aipleasetellme.com.
Written by the AI, TELL ME! team (aipleasetellme.com), November 2026.


