Which sources do AI engines cite when recommending SEO software?

An ongoing study of the third-party domains ChatGPT, Claude and Perplexity cite when answering non-brand questions about SEO and AI-visibility software.

Methodology

1A fixed set of non-brand prompts ("best SEO software", "best AI visibility tools", "X alternative", "SEO software for agencies", and similar) is run against ChatGPT (OpenAI API), Claude (Anthropic API) and Perplexity, all with live web search.
2For every answer we record the brands recommended and every cited URL and domain.
3Cited domains are aggregated across prompts and engines: how often each is cited, how many distinct prompts it feeds, and which brands it carries.
4Runs are repeated on a schedule so the source pattern can be tracked over time rather than reported as a one-off snapshot.
5The same pipeline that powers Semantyra’s AI Authority Gap feature is used, so the method is reproducible by any customer on their own category.

Sample

Collection in progress

Findings

Data collection is in progress. The first findings publish here once the sample is large enough to report responsibly.

Limitations

AI answers vary by phrasing, session and time; the fixed prompt set and repeated runs reduce but do not remove this.
Coverage is limited to ChatGPT, Claude and Perplexity. Google AI Overviews and Bing Copilot are not included because there is no reliable, compliant way to measure their citations yet.
A domain being cited frequently does not prove it causes a recommendation; it is a correlation worth acting on, not a guarantee.

Dataset

A downloadable dataset will be published alongside the first findings.

Citing this work

Semantyra. “Which sources do AI engines cite when recommending SEO software?”. 2026. https://semantyra.com/research/ai-search-citation-study-2026

Author: Semantyra research team. Contact: [email protected].