What are AI search’s hidden competitors? They are the domains AI engines cite that never rank in Google’s top 10 for the same queries. I am Andrii Byzov, an AI-native fractional CMO and GEO consultant for B2B SaaS, and in my GEO vs SEO comparison of 148 real buyer queries across 28 software categories I found a clear split: 343 of the 508 domains AI cited never appeared in Google’s top 10, while only 165 domains overlapped both surfaces. If you track only Google rankings, that 343-domain set is invisible to you, even though it is competing for the answers buyers read.
Key takeaways
- For the same 148 queries, 451 domains ranked in Google top 10, 508 were cited by AI, and only 165 overlap.
- That leaves 343 AI-cited domains (about 68%) that never appear in Google’s top 10 for these queries. This is the hidden competitive set.
- The most-cited AI-only domains here were ahrefs.com, mixpanel.com and semrush.com (7 queries each), then helpscout.com and intercom.com (6 each).
- An SEO-only monitoring stack cannot see these domains, because rank trackers watch Google organic, not AI citations.
- Caveats up front: this is correlational, not causal, measured on 148 US queries, and high citation count is visibility, not proven quality or trust. It does not cover Google AI Overview sources, which are not in this data.
This does not mean Google rankings stopped mattering. It means a Google-only view of the competitive landscape leaves most of the AI surface unmonitored.
The hidden set, defined
When a B2B buyer asks ChatGPT or Perplexity “what is the best X for Y”, the engine names sources. Some of those names also rank in Google’s top 10 for that query, and your rank tracker would catch them. Most do not. Across these 148 queries, 508 unique domains earned at least one AI citation, but only 165 of them also sat in Google’s top 10. The remaining 343 are AI-only: present in the answer, absent from the page you monitor.
That is the practical meaning of a hidden competitor. It is not hidden from buyers, who see it cited in plain text. It is hidden from you, specifically from any monitoring built on Google rankings alone. A competitor can be shaping the AI answer to your category question while your SEO dashboard shows no movement at all, because the dashboard was never looking at that surface.
The most-cited domains you would not be tracking
Here are the AI-only domains cited across the most queries in this sample, the ones that surfaced repeatedly in AI answers yet did not reach Google’s top 10 for those same queries. Read the count as how many of the 148 queries each domain appeared in, which is a visibility signal in this sample, not a quality or trust ranking.
| Rank | AI-only domain | Queries cited (of 148) |
|---|---|---|
| 1 | ahrefs.com | 7 |
| 2 | mixpanel.com | 7 |
| 3 | semrush.com | 7 |
| 4 | helpscout.com | 6 |
| 5 | intercom.com | 6 |
| 6 | toolradar.com | 5 |
| 7 | techrepublic.com | 5 |
| 8 | notion.so | 5 |
| 9 | activecampaign.com | 5 |
| 10 | mailerlite.com | 5 |
| 11 | surferseo.com | 5 |
| 12 | jasper.ai | 5 |
| 13 | copy.ai | 5 |
| 14 | aws.amazon.com | 5 |
Look at the names. Most are established category brands: analytics, support, email, SEO tooling, AI writing, cloud. A few, like toolradar.com and techrepublic.com, are aggregators and editorial sites that AI leans on as references. None of these reached Google’s top 10 for the queries where AI cited them, in this sample. So a team watching Google would not have these domains on any list, yet AI treated them as worth quoting back to a buyer.
I want to be careful about what the count means. A 7 here means the domain showed up in seven of the 148 queries. It does not prove the domain is more trusted, more accurate or higher quality than a domain with a 4. It is a frequency-of-appearance signal within one sample. Treat the table as a map of where AI attention concentrated, not a leaderboard of merit.
Why SEO-only monitoring misses them
The mechanics are simple. A rank tracker queries Google and records which domains occupy the top 10 for your terms. That is its entire field of view. AI citations are produced by a different system that scores different things, and in this sample the two lists overlapped on only 165 of 508 AI-cited domains. So roughly two-thirds of what AI cited has, by construction, no row in a Google-only tool.
This is also why some vendors feel invisible in AI despite ranking well, and why others get cited without obvious top rankings. The systems are not measuring the same thing, so a strong organic position does not reliably carry into AI answers, and a weak organic position does not rule a brand out of them. Directionally, AI engines appear to reach for recognizable category entities as references even when those entities are not ranking organically for the exact phrasing. That favours broad, corroborated entity authority over single-query on-page tuning, though this data shows the pattern, it does not prove the cause.
What to do about the hidden set
Three moves follow from the data, in this sample. First, monitor the AI surface separately. If 343 of 508 cited domains never touch Google’s top 10, then a Google-only competitive view is missing most of the AI picture by design. Second, build a real AI-citation watchlist: track who AI names for your category questions, not just who ranks. Third, invest in entity authority, because AI cited strong category brands even when they were not ranking, which suggests credible, corroborated presence across the web earns citations that a single-query optimization does not.
This is the work I do as a GEO consultant. If you want the surface measured and a watchlist built, GEO is the service, and AI Visibility Score is the diagnostic that shows where you and your hidden competitors stand in AI answers today. For the full underlying study, see my AI Search Visibility Benchmark, and for how AI citations and Google rankings diverge, what AI cites vs what Google ranks covers the overlap in detail. I also write more about GEO on LinkedIn.
Honest limits
This is a directional snapshot, not a law. The numbers are correlational, not causal: I can show that 343 AI-cited domains did not rank in Google’s top 10, but not why any single one was or was not cited. “Never in top 10” means within this sample of 148 US-context queries, on the dates captured, for Google’s top 10 organic results only, excluding ads, maps and other result types. The AI side covers ChatGPT and Perplexity and can shift by account, region and phrasing, so counts will move when I re-run it. Citation counts are visibility within this sample, not proven quality or trust. And this analysis does not include Google AI Overview sources, which were not part of this data, so nothing here should be read as a claim about AI Overviews. Treat the hidden set as a signal to widen what you monitor, not as a finished competitive ranking.
FAQ
What are hidden competitors in AI search? Domains AI cites when answering buyer questions but that never appear in Google’s top 10 for the same queries. In this sample of 148 B2B SaaS queries, 343 of 508 AI-cited domains, about 68%, never ranked in Google’s top 10. A Google-only monitoring setup cannot see them.
Which domains does AI cite most that Google does not rank? In this sample the top AI-only domains by query count were ahrefs.com, mixpanel.com and semrush.com (7 each), then helpscout.com and intercom.com (6 each), and toolradar.com, techrepublic.com, notion.so, activecampaign.com, mailerlite.com, surferseo.com, jasper.ai, copy.ai and aws.amazon.com (5 each). Counts are visibility, not quality.
Why do SEO-only tools miss these competitors? Rank trackers watch Google organic results. They do not see AI citations. Since 343 of 508 AI-cited domains here never reached the top 10, a Google-only setup has no row for them. This is correlational and measured on 148 US queries, but directionally a whole competitive set can be active in AI answers while your dashboards show nothing.
Does a high AI citation count mean a domain is trusted? No. Citation count is how often a domain appeared across this sample, which is a visibility signal, not a quality or trust score. The snapshot is correlational, not causal, “never in top 10” means within these 148 US queries, and it does not cover Google AI Overview sources.