Bevizio
← Blog

How to Rank in ChatGPT: What Is Actually Known

By Bevizio · 30 September 2026

We found no reliable way to make ChatGPT name your product. The evidence falls into four kinds: access rules that companies document, correlations, the main controlled experiment we found, and an audit by a vendor. None of them is a recipe.

What companies document

This is about access, and it is the firmest ground. OpenAI says OAI-SearchBot is used to surface websites in ChatGPT’s search features, and recommends allowing it in your robots.txt file. For Google, a page must be “indexed and eligible to be shown in Google Search with a snippet” to appear as a supporting link in AI Overviews or AI Mode. Google adds that there are “no additional requirements” and no “special optimizations necessary”. Being reachable is a condition of being named. It is not a way of being named. More detail is in why AI engines name your competitor instead of you.

What correlates

In a study published on 12 December 2025, Ahrefs compared 75,000 brands with a Domain Rating above 40 against how often AI answers mentioned them. Ahrefs is an SEO software vendor, and the data came from its own brand-tracking product. Branded web mentions correlated at 0.664 for ChatGPT, 0.709 for AI Mode and 0.656 for AI Overviews. YouTube mentions correlated slightly higher, at about 0.71 to 0.74. Backlinks and domain rating correlated far less, at about 0.2 to 0.3. Ahrefs states that “correlation isn’t causation”, and the study doesn’t show that getting more mentions makes an engine name you.

The main controlled experiment

Researchers behind a paper presented at KDD 2024 built a benchmark of 10,000 queries across 25 domains and tested nine ways of rewriting a source page. Their main experiments used a simulated engine built on gpt-3.5-turbo with the top five Google results, and they checked some of the methods on Perplexity.ai using 200 queries. The best three methods were adding quotations, adding statistics and citing sources. On one of their two measures, how much of an answer was attributed to the page, with earlier mentions weighted more, these raised the page’s visibility by roughly 30% to 40% relative to the baseline, by the authors’ account. Keyword stuffing scored lower than the baseline on that measure.

The limits matter. It measured whether a page was used and cited inside an answer, not whether a brand was recommended. The main experiment’s model is from 2023, the results are a relative gain on the authors’ own metric, and the authors note the methods may need to change as engines change.

What a vendor audit adds

A May 2026 preprint from Unusual.ai, which sells AI brand marketing, audited about 37,000 runs and found that about half of the brands in its two lowest-profile tiers never surfaced. For the smallest startup tier the authors say the data implies three channels: authority lists, canonical content on platforms their search services reach, and inclusion in third-party comparison articles. They did not test whether those work, and the paper has not been peer reviewed as far as its arXiv page shows. Do AI engines ignore small software? covers it in full.

What nobody has shown

We found no study that changed something about a real brand and measured, across repeated answers, whether an engine then named it more often. The closest we found is a June 2026 preprint on one site, glasp.co, which changed some of its pages and measured ChatGPT referral traffic, not naming. It estimated a 1.82 times increase (95% confidence interval 1.31 to 2.54), but a placebo test on the same data came out at p=0.16, and the authors call the result “suggestive, not conclusive”. That is a gap in the evidence, not proof that nothing works. If you try something, the only way to know is to count how often you are named before and after, across many answers. One answer can’t tell you, as why we ask the same question 8 times a week explains. The AI visibility explainer covers how the counts work, and the how-to-check guide covers doing it by hand. The free check runs one real question across ChatGPT, Perplexity and Google AI Overviews with no account.

Drafted with AI assistance. Every factual claim was checked against the sources above.

See whether ChatGPT, Perplexity and Google AI Overviews name your product. No account needed.

Run a free check