Who we are
Three Impact is published by Space & Story, a small web studio in Toronto, and by the team that builds cite-met, a hosting and AI analytics platform for small-business websites. It's written by Mahmoud Halat, who founded Space & Story and spends most of his time on small-business sites and how they get found, which these days mostly means arguing with robots.txt files.
The name comes from the three places a small business now gets found: Google, AI answers, and its own website. Most writing about the middle one is advice. We wanted to publish measurements, because we happen to be able to.
Who owns what
We want this to be impossible to miss, so here it is plainly. We own and operate cite-met (cite-met.com), and cite-met appears in our comparisons. Wherever it does, the page carries a disclosure box at the top and says so again in the text. We also give it real weaknesses, and we name the competitors that are a better fit for most readers when that's true, which on our tools comparison it often is.
How we make money: Three Impact doesn't take affiliate commissions, sponsored placements or payments from any vendor we write about. It's funded by Space & Story and cite-met, the products we've just told you we own. There are no ads.
Where the Observatory data comes from
The AI Search Observatory is built from cite-met's server-side analytics: daily counts of requests per site, per bot and per traffic channel, classified by user agent. For each edition we run read-only queries against those rollups and freeze the results into the published page, with the date we pulled them.
What we include. Small-business sites hosted on cite-met with a verified custom domain and at least 30 days of data. What we leave out. Sites run by our own team, demos and previews, non-business apps, sites cite-met only monitors through a beacon (which can't see crawlers), and any site large enough to dominate the totals on its own.
How we protect the sites in the sample
The sites in our data belong to cite-met customers, and their trust matters more to us than a better chart. So we follow fixed rules:
- We publish aggregates only. No site, client, industry or location is named or hinted at, and no per-site number appears anywhere.
- Every figure covers at least five sites. Any bot, bucket or breakdown seen on fewer is merged into "others" or left out. If the sample is too small for a breakdown at that threshold, we publish the total only.
- Shares are site-weighted, so no single site can drive a result.
- We never write to the production database, and the queries are read-only.
How we research everything else
- We open the source. Every user agent, robots.txt token, price and quoted statement on this site was checked against the publisher's own page in October 2026 and links to it. Where an operator publishes nothing, we say so instead of copying what other lists guess.
- We say what we did. When we compared published specs, we say "we compared". We only say "we tested" or "we measured" when we did, and the Observatory is the only measured data on the site this quarter.
- We flag weak sources. When a claim rests on a third-party write-up rather than the company's own documentation, the table says so.
- We update quarterly. Each Observatory edition comes with a recheck of the registry and the price tables. The "Updated" date at the top of each page is the last time we checked it.
The tools
The AI access checker fetches three files from the domain you enter (robots.txt, the homepage and llms.txt) from our server, with a five-second timeout and a 512 KB limit, and doesn't store the domain or the results. It's free and needs no sign-up.
Corrections and contact
If you find something wrong, or you run one of the crawlers or tools we list and our description is out of date, tell us and we'll fix it and note the change. The fastest route is through Space & Story. We'd rather be corrected than be confidently wrong in a table people copy.