How to Get Your Business Mentioned by AI Search Fetchable, Readable, Corroborated
Somebody in your town is asking an assistant for a plumber, a dentist, a driving instructor or a wedding photographer right now, and getting three names back. There is no ranking report to check and no place to submit your site. But the process is less mysterious than it looks, and it comes down to three conditions that you can each verify yourself in an afternoon.
Fetchable, readable, corroborated
An assistant answering "best X near me" does roughly this: it assembles a shortlist of candidate businesses from sources it already trusts, then fetches some pages to check details, then writes an answer citing what it read. Three things can stop you, and they fail in a specific order.
- Fetchable. If the crawler cannot retrieve your page, nothing else matters.
- Readable. If it retrieves a page whose facts are painted by JavaScript, it learns nothing from you.
- Corroborated. If nothing outside your own domain says you exist and what you are good at, you rarely make the shortlist in the first place.
Most businesses worry about the third one and have quietly failed the first two.
1. Fetchable: read your own robots.txt today
Open yoursite.com/robots.txt in a browser. You are looking for Disallow rules against the AI crawlers, which each vendor runs under more than one name. OpenAI uses GPTBot for training and OAI-SearchBot for search, plus ChatGPT-User when a person asks it to open a link. Anthropic uses ClaudeBot and Claude-SearchBot. Perplexity uses PerplexityBot and Perplexity-User. Google separates Google-Extended from Googlebot. Microsoft uses bingbot for the index that Copilot draws on, and there are others: Applebot, Amazonbot, CCBot, YouBot, Bytespider, Meta-ExternalAgent.
The common accident is not a decision you made. Several hosts, site builders and CDNs turned on AI bot blocking by default, so a site can be closed to assistants without anyone choosing it. If you find rules you did not write, that is your answer, and removing them is a five-minute job. Decide deliberately either way: blocking is a legitimate choice, it just should be a choice.
Second half of fetchable: be in both indexes. Copilot and ChatGPT search retrieve heavily from Bing, while Gemini and Google AI features retrieve from Google. Plenty of sites are healthy in one and thin in the other without ever noticing, because almost nobody opens Bing Webmaster Tools. Both consoles are free. Submit your sitemap to both.
2. Readable: the facts have to be in the HTML
AI crawlers take the HTML your server sends and stop. They do not run your scripts and they do not read the contents of a third-party iframe as part of your page. So anything that only appears after JavaScript runs is invisible to them, however prominent it looks in a browser.
On a typical small business site that quietly hides several of the most important facts: the opening hours in a booking widget, the price list in an embedded calculator, the service area in a map, the testimonials in a slider, and the reviews in a review widget. Check them the same way each time: view-source on the page, then search for the exact words.
What to put in plain text, at minimum: business name, what you do in the words customers use, the areas you serve, opening hours, prices or a price range, and your review rating with a count. Add Organization or LocalBusiness structured data on top, which helps machines resolve you to an entity rather than a string. An llms.txt file is cheap to publish and gives assistants a plain-text map of your important pages.
Reviews are the piece most often missing, because almost every review widget renders in an iframe or by script. The two-minute test tells you where you stand, and the platform-by-platform guide covers what Facebook, Trustpilot, Angi and the rest actually leave behind. On BlooTrue paid plans the embed writes review text, rating and count into your page HTML with schema.org markup and then swaps in the animated widget for visitors; free BlooTrue widgets render in an iframe that crawlers skip.
3. Corroborated: your own website is not enough on its own
This is the part that surprises people, and it is the reason a beautiful, fast, perfectly-marked-up website can still be absent from every answer. Assistants do not begin by reading your site. They begin by working out which businesses are candidates, and that shortlist is built from places other than your domain: review platforms and directories, maps and structured listings, comparison articles and roundups, forum and community threads, and industry associations.
If none of those name you, you never enter the conversation, and no amount of on-site work changes that. The corroboration list for a local business is unglamorous and mostly free:
- A complete Google Business Profile, with hours, services, photos and categories filled in properly.
- The review platforms your industry actually uses, claimed and complete, not just Google.
- Consistent name, address and phone number everywhere they appear, because inconsistency reads as two different businesses.
- Real answers to real questions in public places where your customers already ask them, written as yourself and disclosing who you are.
- Being listed in the local, industry and association directories that already rank for your category.
None of it is new advice. What is new is that the payoff arrives faster than it used to, because an assistant reading three sources that mention you can act on that immediately, where a search engine used to need a lot more signal.
Do you need a different plan for Claude, Gemini and Perplexity?
No, and it is worth saying plainly because the opposite is widely implied. The three conditions above are the same for all of them, because they all read HTML rather than rendered pages, and they all build a candidate list before they read anything of yours. Work done for one is done for all of them.
They differ in exactly two ways, both already covered. They retrieve from different indexes, so being in Google and Bing covers most of the field. And they send different crawler names, which only matters when you are writing robots.txt or a CDN rule. There is no Perplexity markup and no Gemini tag to add.
How to tell whether any of this is working
There is no rank tracker for assistants, so measure three cheaper things instead. Ask the assistants directly, once a month, the questions a customer would actually type, and write down which businesses come back; that is your competitive set. Watch your server logs or analytics for the AI crawler user agents, which tells you whether you are being fetched at all. And watch for referral traffic from chatgpt.com, perplexity.ai and copilot.microsoft.com in your analytics, which is the only part that shows up as real visits.
Expect the numbers to be small and the visitors to be unusually well qualified. Somebody who arrives after an assistant recommended you by name has already been sold.
The one-afternoon checklist
- Open yoursite.com/robots.txt and check for AI crawler blocks you did not add.
- Confirm your CDN or host is not blocking AI bots by default.
- Submit your sitemap to Google Search Console and Bing Webmaster Tools.
- View-source and search for your hours, prices, service area and a review sentence.
- Add Organization or LocalBusiness structured data if it is missing.
- Publish an llms.txt listing your important pages.
- Complete your Google Business Profile and the review platforms your industry uses.
- Get your reviews into the page HTML rather than only inside a widget.
- Ask five assistants a customer question and write down who they name.
Being readable is the requirement, not a guarantee. Whether an assistant names your business is its decision, and nobody can sell you a place in an answer. What you can do is make sure that when it looks, there is something there to find.
