HR Software AI Shortlists Split by Engine Reach: 6 of the Top 10 Are Not Universal Sources
The first published HR software buyer-choice leaderboard looks stable at the category layer and fragmented at the engine layer. Six of its top ten sources are absent from at least one engine's citation set, while the HR news list is almost universal.
The first published HR software buyer-choice leaderboard does not mean every AI engine is reading the same shortlist.
In the Machine Relations Index release generated September 21, 2026 (mri_score_v2.0+2026-09-21+d73e239e7c5b), the HR & Talent / how-buyers-choose segment crossed the evidence floor: 101 observed answer runs across seven run dates, with 235 cited source domains. That gives the category its first citable buyer-choice list.
The top ten is already published. The engine layer is less settled. Six of those ten domains are absent from at least one of the six engines' citation sets across the full Index window. Three of the top ten are cited by only two engines in the release overall. Google AI Overviews is missing five of the ten. ChatGPT is missing four. Perplexity is the only engine present on every top-ten domain.
That does not say which engine cited which source inside this exact HR segment. The public release does not expose per-engine rates at segment level. It says something narrower and still useful: the first HR buyer-choice shortlist is made of sources whose whole-window engine reach is uneven.
The buyer-choice top ten, with engine reach
The HR & Talent / how-buyers-choose denominator is 101 observed answer runs. A domain's segment rate is the share of those runs in which the domain was cited at least once. Engine reach below is whole-window presence in the release: whether each engine cited that domain anywhere in the measured panel between May 10 and September 21, 2026.
| Rank | Source | Segment runs | Source role | Whole-window engines present | Missing engines |
|---|---|---|---|---|---|
| 1 | Pin | 14 of 101 | Vendor-owned source | 5 of 6 | Google AI Overviews |
| 2 | 13 of 101 | Community and social platform | 6 of 6 | — | |
| 3 | 9 of 101 | Community and social platform | 4 of 6 | ChatGPT, Claude | |
| 4 | Selenios | 9 of 101 | Other observed source | 2 of 6 | ChatGPT, Gemini, Google AI Mode, Google AI Overviews |
| 5 | Umantis | 9 of 101 | Other observed source | 2 of 6 | ChatGPT, Gemini, Google AI Mode, Google AI Overviews |
| 6 | US Tech Automations | 9 of 101 | Other observed source | 6 of 6 | — |
| 7 | ClearCompany | 8 of 101 | Other observed source | 6 of 6 | — |
| 8 | CVViZ | 8 of 101 | Other observed source | 2 of 6 | ChatGPT, Gemini, Google AI Mode, Google AI Overviews |
| 9 | Gem | 8 of 101 | Vendor-owned source | 5 of 6 | Google AI Overviews |
| 10 | Lever | 8 of 101 | Other observed source | 6 of 6 | — |
Four of the ten are cited by all six engines somewhere in the full release. Six are not. Three sources that sit inside the first eight ranks — Selenios, Umantis and CVViZ — are cited by only Claude and Perplexity anywhere in the window. They are visible enough to enter the first HR buyer leaderboard, but not broad enough to be treated as universal engine infrastructure.
That is the engine-reach warning. A source can rank inside a category answer while still being unevenly distributed across engines.
The news list is broader
The HR news-shaped list looks different. It has 608 observed runs across 53 run dates, so it is thicker, older and less fragile than the new buyer-choice segment.
Its top ten are Unleash, Landbase, Crunchbase, Rippling, SHRM, Outsail, HR Executive, People Managing People, HR Tech Feed and LinkedIn.
Eight of those ten are cited by all six engines somewhere in the full release. The other two are cited by five of six. None is collecting-confidence overall. None is limited to two engines.
| HR & Talent top-ten list | Top-ten domains cited by all six engines | Top-ten domains cited by two or fewer engines | Collecting-confidence domains in top ten |
|---|---|---|---|
| Buyer-choice / how buyers choose | 4 of 10 | 3 of 10 | 3 of 10 |
| News-driven citations | 8 of 10 | 0 of 10 | 0 of 10 |
The difference is not just editorial versus vendor-owned. It is maturity and distribution. HR news answers are drawing from sources with broad engine reach. HR buyer-choice answers just became citable, and the first list is still full of narrower surfaces.
That is why the category read has to keep two layers separate. The buyer leaderboard tells you what sources the model cites for the buyer-choice question. Engine reach tells you whether those sources are visible across the panel or concentrated in a smaller set of engines.
The engine pattern
Across the buyer-choice top ten, missing-engine counts look like this:
| Engine | Top-ten HR buyer-choice sources absent from that engine's whole-window citation set |
|---|---|
| Google AI Overviews | 5 of 10 |
| ChatGPT | 4 of 10 |
| Gemini | 3 of 10 |
| Google AI Mode | 3 of 10 |
| Claude | 1 of 10 |
| Perplexity | 0 of 10 |
This is not a quality ranking of the engines. It is a distribution map.
Perplexity has the widest source layer in the Index overall, so it is not surprising that it is present on every top-ten HR buyer-choice source. Google AI Overviews is a search feature with a narrower citation surface, so it is also not surprising that it misses half of a newly published buyer-choice top ten. ChatGPT's four absences are concentrated in the same lightly classified operator surfaces that also miss the Google surfaces.
The practical consequence is simple: if you only test one engine, you can mistake a source's engine-specific visibility for category authority.
A vendor whose strongest citations come from LinkedIn, Lever and ClearCompany is in a different position from a vendor whose strongest citations come from Selenios, Umantis and CVViZ. The first set is visible across the full panel. The second set is winning a category shape while still narrow at the engine layer.
What this changes for HR software operators
The first HR buyer-choice leaderboard already killed one lazy assumption: the buyer shortlist is not an HR trade-press list.
The engine layer kills a second one: a top-ten source is not automatically a six-engine source.
For an HR software company, that changes the diagnostic sequence.
First, read the category leaderboard to see what kind of source the answer engines use for the question. In this case, the answer is vendor pages, community platforms and operator surfaces, not editorial publications.
Second, read engine reach before calling any source a durable citation surface. LinkedIn, ClearCompany, US Tech Automations and Lever are all-six-engine sources in the full release. Selenios, Umantis and CVViZ are not.
Third, separate news strategy from buyer-choice strategy. The HR news layer is much more broadly distributed across engines and contains the trade press. The buyer-choice layer is younger and narrower. Treating those as the same source market would erase the only useful signal.
What to watch next
The next useful read is not another top-ten recap. It is a movement test.
Best-tools, is-it-worth-it and problem-first are each one run date short of the HR evidence floor. When one of them publishes, the question is whether the same narrow sources carry across, or whether the top ten shifts toward all-six-engine domains.
Three outcomes would matter.
If Selenios, Umantis, CVViZ or another two-engine domain appears again, the HR buyer layer may be forming around narrow but repeated operator surfaces.
If People Managing People or SelectSoftware Reviews moves into a buyer top ten, the trade press is shape-sensitive rather than absent.
If the next buyer shape shifts toward all-six-engine sources, then this first HR buyer-choice list is a first-published slice, not a durable category pattern.
Until then, the clean reading is narrow: HR buyer-choice citations have a first leaderboard, and that leaderboard is more engine-fragmented than the HR news list.
Method note
Source: the public Machine Relations Index release mri_score_v2.0+2026-09-21+d73e239e7c5b, generated 2026-09-21, window May 10 through September 21, 2026, with 16,146 observed answer runs, 126,924 citation events and 22,377 cited domains across ChatGPT, Claude, Gemini, Google AI Mode, Google AI Overviews and Perplexity.
The HR & Talent buyer-choice figures are read from the public HR & Talent / how buyers choose leaderboard. The HR news comparison is read from the HR & Talent / news-driven citations leaderboard. Domain-wide engine reach is read from each domain's public Machine Relations profile under /index/domains/<domain> and means the engine cited that domain at least once anywhere in the full measured panel. It is not a per-segment engine rate, and this article does not infer which engine cited which source inside the HR buyer-choice segment.
FAQ
Do AI engines cite the same HR software sources? Not in a useful sense. In the first published HR buyer-choice top ten, only four of ten domains are cited by all six engines somewhere in the full release; three are cited by only two engines.
Which engine reaches the whole HR buyer-choice top ten? Perplexity is present on all ten domains in the full release. Google AI Overviews is absent from five, ChatGPT from four, Gemini and Google AI Mode from three, and Claude from one.
Does this prove Perplexity cited every source inside HR buyer-choice answers? No. The public release exposes whole-window engine presence by domain, not per-engine rates inside this segment. The claim is about source reach across the release, not per-segment attribution by engine.