Nobody outside OpenAI can hand you a confirmed formula for how ChatGPT chooses which local business to name. What follows is a set of practical steps. They line up with how these systems appear to gather and cite information, based on what shows up in their answers and what a page needs before a crawler can read it at all. Treat this as informed and hedged, not a guarantee.
Start With Whether the System Can Read Your Site at All
Before anything else can matter, your site has to be crawlable. Check your robots.txt file. Confirm you are not blocking crawlers your business actually wants visiting. Check something else too. Do your service details, prices, and descriptions load only after JavaScript runs? A crawler that reads only the raw page source can miss content that a human browser renders fine. To check, view your page source directly rather than the rendered version, and confirm your key details appear in that raw output.
This groundwork does not guarantee a mention. It removes a basic, avoidable reason your business might be invisible to these systems, whatever else you do right.
Add Structured Data So Machines Can Read Facts Cleanly
Structured data, sometimes called schema markup, is a standard way of labeling facts on a page. Things like your services, your location, your hours, and your review rating, so software can read them directly rather than guessing from the surrounding text and design. Adding LocalBusiness and Review schema to your pages appears to give AI systems a more dependable set of facts to work from, compared with a page where that same information sits only inside styled text or a graphic. No confirmed ranking signal is tied to schema markup for AI answers. It is a low-cost, well-documented step that helps traditional search too, so there is little reason to skip it.
Build Reviews Because the Text Itself Gets Read
Picture a customer typing a narrow, specific question into ChatGPT, checking a plumber's punctuality or asking how a dentist handles a nervous patient. In cases like that, the assistant appears to draw on the written body of your reviews rather than stopping at a plain star average. A business backed by real, detailed reviews across Google, Facebook, and Bing hands these tools far more usable material than one that shows only a number next to its name. Review text sits among the more controllable levers here. A business can grow it steadily through an ordinary request process, rather than waiting on an outside algorithm.
Earn a Mention on Pages These Systems Already Read
Comparison guides, local roundups, and third-party review sites carry weight in how these systems decide who to name, judging by how often their citations trace back to exactly that kind of writing. Landing coverage there takes ordinary outreach. You pitch relevant publishers and roundup writers honestly and give them a real reason to include your business. No trick manufactures the same result overnight.
The llms.txt Debate: Two Sides, No Settled Answer
A file called llms.txt has been proposed as a way for a site to hand AI systems a clean, plain-language summary of what it offers, separate from the messier full HTML a crawler otherwise has to parse. Supporters argue it gives these systems an easier, more accurate source to work from, similar to how a sitemap helps traditional search engines find pages. Skeptics point out that no major AI system has confirmed it reads or weighs llms.txt files in a way that changes what gets cited. Publishing one costs almost nothing, which makes it easy to add without proving it helps.
The honest position sits between those two views. Adding an llms.txt file is low effort and unlikely to cause harm. No confirmed ranking signal is tied to it today, so it belongs near the bottom of a priority list rather than treated as a fix on its own.
Check What These Systems Currently Say About You
Set aside a few minutes on a recurring basis. Put yourself in your customer's seat and type the exact phrasing they would use into ChatGPT, Perplexity, and Google, the kind of question a stranger types when hunting for a local plumber. Read carefully whether your business comes up by name and which pages get cited as the source. Ask the same thing more than once in one sitting, since the wording and even the named businesses can shift between attempts. One answer alone does not tell you much. A monthly rhythm for this check tends to be enough to spot a genuine change without overreacting to ordinary noise between one attempt and the next.
Keep the Facts Matching Everywhere They Show Up
A detail worth building toward once the basics above are in place. Your services, prices, service area, hours, and contact details should read the same on your own site as they do on any roundup, directory, or review platform that mentions you. Mismatched facts in different places give an assistant conflicting inputs to reconcile. When that happens it may lean toward whichever source it treats as more current or authoritative, which is not guaranteed to be yours. Keeping the same facts everywhere is unglamorous work. It removes a source of doubt these tools would otherwise have to resolve on their own.
What This Adds Up To
None of the steps above come with a guarantee. Any claim of a confirmed formula for AI recommendations deserves real skepticism, since these tools update constantly and their exact process stays hidden from everyone outside the companies that build them. Taken together, these steps leave a business better positioned. A crawlable site, clean structured data, real review text spread across platforms, and honest coverage from third parties. Building those things pays off on its own terms, whether or not a specific ChatGPT mention ever shows up as proof.
If you want a hand keeping track of what these systems currently say about your business and working on the pieces behind it, see our AI search visibility service.