Short answer: we registered aivisibilitycensus.com on 29 October 2026 and launched it the same day with a crawler log that verifies bots against published IP ranges, Markdown versions of every page, llms.txt, IndexNow and Content Signals. Bingbot made a verified visit the same morning. Below is the full setup, what the logs showed by mid-afternoon, and four predictions written before the results come in.
Why we are doing this
Most advice about getting found in AI search is untested. This site gives us a clean test bed: a domain with no history, no links and no traffic. Every crawl, index and citation from here on is something we caused or something we can watch arrive. We will publish all of it, including the results that show nothing.
The setup
Hosting and platform. WordPress 7.1 on GoDaddy shared hosting, behind Cloudflare’s free plan. A custom block theme, no page builder and no SEO plugin. Everything an SEO plugin would do (titles, meta descriptions, canonical tags, robots rules, structured data) is written into the theme by hand, so we know exactly what each page sends.
Two versions of every page. Every page has a Markdown version at the same address with .md on the end. HTML pages point to it with a Link: rel="alternate"; type="text/markdown" header, and any request that asks for Markdown in its Accept header gets it. The content is identical to the HTML. Bots get no special version of anything.
llms.txt and llms-full.txt. Generated automatically from every published page and post, using each page’s summary.
IndexNow. Every time a page is published or updated, its HTML address, its Markdown address and llms.txt are submitted to IndexNow, which shares them with Bing and the other search engines that take part. Each submission and its response code is logged.
Content Signals. Our robots.txt says Content-Signal: search=yes, ai-input=yes, ai-train=yes: yes to search, yes to use in AI answers, yes to AI training.
Structured data. One connected graph on every page: Organization, WebSite, the page itself and breadcrumbs, plus Article and FAQPage on articles.
A crawler log we can trust. Every bot request is logged, plus every request for a Markdown page or llms.txt. Where a company publishes its crawler IP ranges (OpenAI, Perplexity, Google, Microsoft, Apple), each visit is checked against them and marked verified or spoofed. Because the site sits behind Cloudflare, the log only trusts the visitor IP that Cloudflare passes on when the request really came from Cloudflare.
A time-to-first-crawl table. For every published page: hours from publishing to the first visit from a search engine crawler, an AI search crawler, an AI assistant fetching a page for a user, and an AI training crawler.
Day one, in numbers
Log figures below are as of 14:10 UTC on 29 October 2026.
| What | When or how many |
|---|---|
| Domain registered | 29 October 2026 |
| Census pages published | From about 10:20 UTC. A default WordPress install was already live on the domain before that. |
| First verified crawler | Bingbot, 10:43 UTC. Its only visit so far. |
| First article published | 11:27:17 UTC, with IndexNow submitted one second later |
| IndexNow | The five submissions made between 11:04 and 11:28 UTC were all answered with response code 200 |
| YandexBot | 7 requests from 12:09 UTC, all for robots.txt and sitemap files. Yandex takes part in IndexNow. |
| ClaudeBot | 10 requests from 11:10 UTC, mostly robots.txt and sitemap files. Unverified: we have no published IP list to check them against. |
| AhrefsBot | 29 requests from 12:17 UTC, including llms.txt and the Markdown version of every page it visited |
| Googlebot, GPTBot, OAI-SearchBot, PerplexityBot | None yet |
| Cloudflare Agent Readiness scan | 4 of 5 quick wins, 1 of 3 technical groundwork checks, 0 of 8 advanced integration checks. The missing quick win was Content Signals, which we then added. |
The first crawler to ask for llms.txt and our Markdown pages was an SEO tool’s crawler, not an AI company’s. No AI crawler has requested either yet.
What we predict
Written on 29 October 2026, before any of these results exist.
- Bingbot reaches each new article before Googlebot does.
- Our first article’s test marker, “Census test marker CT-001”, shows up in Copilot or Perplexity answers before it shows up in ChatGPT or Google.
- No verified AI crawler requests llms.txt in the first 30 days.
- Markdown versions make up less than 5% of verified AI crawler requests in the first 30 days.
If we are wrong, we will say so here.
What we will not do
We will not show bots different content from what people see, hide text or instructions aimed at AI models, change a page’s date without changing the page, or build links we would not want a reader to see. The full rules are on our experiments page.
What happens next
Three running studies come out of this setup: how long a new site takes to be indexed and ranked in Google, how long it takes to be cited by AI search tools, and which AI search tactics make a measurable difference. All three are updated as results arrive, and every experiment is listed on the experiments page.
About this article
This article is part of the same test. Its test reference is Census test marker LN-001. If an AI tool tells you about LN-001, it has read this page.
Published by the AI Visibility Census. Also available as Markdown.