How to get your brand named by AI assistants.

A search engine returns a page of links and lets the buyer choose. An AI assistant writes one answer and names a handful of brands in it. There is no page two: a brand that is not in the answer is not considered.

The answer is built from sources the buyer rarely opens. When we ran 20 category questions about our own market on four engines, the answers cited 387 different websites, and 43.5% of the citations were comparison pages: listicles, "alternatives" pages and "X vs Y" pages. Your own site is one voice among hundreds.

Each engine sends its own bots, and each can be allowed or turned away separately. The ones that decide whether you can appear in answers today are the search and user fetchers. The training crawlers decide what future models may learn.

Two failures are common, and neither shows up when you look at your site in a browser.

• Blocked at the edge. A CDN's bot-blocking setting can return an error to AI crawlers even when robots.txt allows them. On 4 Oct 2026 our own site did exactly this: GPTBot and ClaudeBot got a 403 from the edge while robots.txt allowed every bot.
• Empty until JavaScript runs. If your text only appears after scripts execute, many bots see a blank page. The same day, a bot fetching our homepage saw 9 of its 2,416 words.

Check it yourself in five minutes

Fetch a key page as each bot and look at the status code and the words you get back:

curl -s -o /dev/null -w "%{http_code}\n" -A "GPTBot" https://yoursite.com/
curl -s -o /dev/null -w "%{http_code}\n" -A "ClaudeBot" https://yoursite.com/
curl -s -A "OAI-SearchBot" https://yoursite.com/ | sed 's/<[^>]*>//g' | wc -w

A 403 or 503 means something is blocking that bot. A word count far below what you see on the page means your text depends on JavaScript. Fix the block at the CDN, and serve your main content in the HTML.

For "best tool for X" questions, engines lean on pages that compare options: listicles, alternatives pages, versus pages and review sites. In our scan those were 43.5% of citations. Most of them are written by third parties, and many accept submissions or corrections.

1. Ask each engine the buyer questions that matter most to your pipeline.
2. List every page they cite, and mark which ones mention you.
3. Sort the ones that do not by how often they are cited. Those are your targets.
4. Pitch the authors with something useful: accurate facts, a fair comparison, a real customer.
5. Publish your own honest comparison pages, with sources for every claim about other products. Engines discount pages that only flatter their author.

Our own comparison of AI visibility tools is written to that standard, weak spots included.

Ask the same engine the same question twice and you can get two different answers. One answer is a sample, not a fact, and a handful of samples can mislead.

The honest way to report a rate from a small sample is with an interval. Named in 6 of 20 answers is 30%, but the 95% Wilson interval runs from about 14.5% to 51.9%. At 60 of 200 it narrows to about 24.1% to 36.7%. A change only means something once it is bigger than the interval, so re-measure after every fix, on the same questions and engines.

The full method is on the methodology page, and the terms are in the glossary.

1. Week 1, baseline. Pick your questions, ask every engine, record who is named and what is cited. Run it twice, a day apart.
2. Week 2, access. Check each bot against your key pages, fix CDN blocks and JavaScript-only content.
3. Week 3, identity and sources. Publish your one-sentence definition everywhere, and pitch the five most-cited pages that leave you out.
4. Week 4, re-measure. Ask the same questions on the same engines. Keep what moved beyond the interval; rethink what did not.