TL;DR: Across 505 tracking runs on 22 buying questions, the engines named the same small group of vendors and cited a much wider set of sources. Semrush was the most cited domain by a wide margin, and most competitor citations came from one question type.
What was measured
Peak Answer ran its own category through the same tracking it runs for customers. That was 22 buying questions about AI visibility and AI SEO software, asked weekly across 5 engines (ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews), with every cited source recorded. The analyzed window has 505 runs, measured in September 2026. These numbers will drift. The vendors themselves are compared in best AI visibility tools in 2026.
The most-cited domains
| Domain | Citations | Mostly driven by |
|---|---|---|
| semrush.com | 345 | use case, pricing, category questions |
| rankability.com | 143 | alternatives questions |
| ahrefs.com | 142 | pricing questions |
| g2.com | 137 | alternatives questions |
| zapier.com | 125 | alternatives, category |
| reddit.com | 118 | alternatives questions |
| otterly.ai | 110 | alternatives questions |
| tryprofound.com | 109 | alternatives, comparison |
| geoptie.com | 106 | alternatives questions |
| frase.io | 89 | alternatives questions |
| peec.ai | 87 | alternatives, comparison |
| cituna.com | 83 | alternatives questions |
569 distinct domains were cited in total. A few sit at the top and the rest is a very long tail.
The engines disagree
| Engine | Searched before answering | Distinct domains cited | Most-cited source |
|---|---|---|---|
| Perplexity | Nearly always | Highest | Varied, heavy on recent articles |
| ChatGPT | Often, varies by question | High | semrush.com |
| Often | Moderate | Established publishers | |
| Claude | Varies most by question type | Moderate | Mixed |
| Gemini | Varies | Lower | Google-shaped |
A brand can be well represented on Perplexity and missing on ChatGPT. A measurement on one engine will give a confident answer that may be wrong for the others.
Which brands get named
Across the 22 questions, in rough order of how often they appeared:
| Brand | Named on | Typically on |
|---|---|---|
| Profound | Most questions | Category, comparison, enterprise framing |
| Peec AI | Most questions | Alternatives, comparison |
| Otterly | Many | Alternatives, cheapest framing |
| AthenaHQ | Many | Agency and multi-brand framing |
| Semrush AI Visibility | Many | Category and pricing questions |
| Writesonic | Some | Content-led framing |
| Scrunch AI | Some | Technical framing |
| Newer entrants | One or two each | Alternatives questions |
The newer entrants are the interesting group. In a category this young, appearing on 2 questions out of 22 is not far from appearing on 5. The gap is a handful of citations.
What drove the citations
| Question pattern | Questions | Citations produced | Sources |
|---|---|---|---|
| alternatives to [rival] | 6 | about 711 | Competitors and comparison sites, almost exclusively |
| use case | 4 | about 243 | Established SEO publishers |
| category | 4 | about 212 | Established publishers and roundups |
| pricing | 2 | about 196 | Publishers with pricing pages, review sites |
| comparison | 4 | about 61 | Competitors' own sites |
| objection | 2 | about 25 | Forums, opinion pieces |
Semrush leads on general questions
Semrush has 345 citations, nearly two and a half times the next domain. It got them from use case, pricing and category questions. For general questions about the category, the engines reach for established SEO publishers. They reach for specialist sites when the question names a specific vendor.
Alternatives questions produced most of the competitor citations
The 6 "alternatives to [rival]" questions made up more than 700 of the roughly 1,450 citations in the table above. Asking an engine for a rival's alternatives gets you a list of competitors, and the sources are competitor comparison pages. Those are not sites you can get a link from.
This is also why rankability.com has two numbers. It has 143 citations in total, and 83 of them came from the alternatives questions. The same applies to the others in that group: geoptie.com has 95 of its 106 from alternatives questions, and cituna.com has 72 of its 83.
Reddit is a top-10 source
Reddit has 118 citations, almost all from alternatives questions. The engines read threads where people compare tools. A small brand can answer a question in one of those threads, and it can't do much about Semrush's domain.
What this means for a brand in this category
General category questions. "Best AI SEO software" is answered out of Semrush, Ahrefs and the large comparison sites. Competing there directly is a long project.
Specific and problem-aware questions. The tail of 569 domains is where smaller sites show up, and they show up on narrower questions.
Forum threads. Reddit is the one source on the list that a small company can get into this month.
What Peak Answer changed
The question set was 20 vendor-selection questions and 2 objection questions. It could only ever return vendors and their comparison pages, so three things changed after this analysis:
- Alternatives questions were cut from 6 to 2. They were worth running once, and they took a quarter of the measurement every week for a list of competitor sources.
- 6 problem-aware questions went into tracking on 30 September 2026, to see what a different question type surfaces.
- Competitors are now inferred from the brands the engines name across different questions, as well as the ones somebody typed in. Six companies the engines kept naming were not on the competitor list, which meant the Links board was suggesting backlink requests to them.
Method and limits
Questions. 22, generated from Search Console data and from what the live search results for this category contain.
Engines. ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews.
Cadence. Weekly, 505 runs in the analyzed window.
Capture. Full answer text, every brand named with its sentiment and position, and every cited source and link, per run. Each citation is attributed to the question that produced it, which is what makes the question-type table possible.
Limits:
- The question set is Peak Answer's own. A different set would give different domains. The overall shape (a concentrated head, a long tail, competitors clustered on alternatives questions) is likely to hold, and the exact ordering is not.
- Citations are unweighted. A domain cited in passing counts the same as one the answer is built on.
- This is one category. The 14 markets in the research studies differ from each other quite a lot, so these numbers won't carry over to yours.
- The figures are from September 2026 and engines change how they retrieve without announcing it.
FAQs
Yes, you can see this for your own category. Peak Answer runs this same method on your questions every week.
The 6 engines disagree with each other, so measuring one only tells you about that one.
These numbers change often enough that every figure here is dated.
505 runs is enough for the overall shape, but not for small differences between neighboring domains. Treat the top 10 as meaningful and the order within it as approximate.