Case study
Launched Oct 25.
Understood by AI by Oct 26.
A brand-new site with no history, no backlinks and a name that sounds like software. One day after launch, four AI assistants described it correctly.
- Site
- crawlist.online, a technical SEO, GEO and AEO consultancy
- Launched
- October 25, 2026
- Checked
- October 26, 2026, about one day later
- Assistants
- Gemini, Claude, Google AI and one more assistant with web search
- Result
- All four named Kean Carl Donato as the owner and described Crawlist as a consultancy, not crawling software
- Test type
- Day-one readability check, prompted with the site URL (see the limits below)
The challenge.
A new domain starts with nothing: no ranking history, no backlinks and no reputation. AI assistants have little to go on, so they lean on what the site itself says and how clearly it says it.
Crawlist had a second problem. “Crawl” sounds like software, and there is a crawling package with a similar name. An assistant that guesses wrong would describe a consultancy as a scraping tool, and a client looking for an SEO specialist would never find it.
What I set up before launch.
- One clear entity, written in codeJSON-LD for the business, the person and the studio, all linked together. It states plainly that Crawlist is a consultancy and not web crawling software.
- A direct answer on the pageA short “What is Crawlist?” block on the homepage, worded to match the meta description and the structured data, so every source tells one story.
- AI crawlers allowed inrobots.txt names the search and retrieval bots, such as OAI-SearchBot, Claude-SearchBot and PerplexityBot. An llms.txt file adds a short machine-readable summary.
- A site that is easy to crawl and indexXML sitemap, a canonical URL on every page, clean URLs, lightweight static pages and Google Search Console verification.
- The same facts on every pageName, role, location and services match across the site, so nothing contradicts the entity definition.
What the assistants said.
Screenshots taken on October 26, 2026, within minutes of each other.
Screenshots are unedited apart from resizing. Wording in the captions is paraphrased.
What this shows, and what it doesn’t.
- Shows
- Four different AI systems read the site and agreed on who runs it, what it offers and what it is not. The name clash did not cause a mix-up.
- Shows
- The structured data, robots.txt and clear wording did their job on day one, before any links or reviews existed.
- Doesn’t show
- Rankings, traffic or leads. In these tests the assistant was given the URL, so this proves the site is readable and well understood, not that it is recommended unprompted.
- Next
- Repeat the checks with no URL in the prompt, track impressions and clicks in Search Console, and update this page with the dated results.
Why this matters for your business.
People ask AI first
More buyers now ask an assistant before they open a search result. If it can’t describe you, you are missing from that conversation.
A wrong answer costs leads
If an assistant mislabels what you do, the right customers never see you, and the wrong ones may arrive.
Early is cheaper
Clear entities and clean markup are far easier to set up at launch than to repair after the web has formed its own picture of you.
It helps Google too
Crawlable pages, canonicals and structured data are the same foundation traditional search rewards, so the work compounds.
Nobody can guarantee a place in AI answers, because they change often. What you control is a technically sound site that says clearly who you are.
Want the same foundation for your site?
I audit how search engines and AI tools read your site today, then fix the biggest gaps first and report in plain language.