How to get your business cited by ChatGPT and Google’s AI

What AI engines actually read

ChatGPT browses through Bing’s index. Google’s AI answers draw on Google’s. Perplexity and Claude fetch pages directly. Most of them do not run JavaScript when they fetch, so a page that renders its content in the browser looks empty to them. The first job is a page whose text is in the HTML the server sends.

The second job is getting into those indexes quickly. Submit your sitemap to Google Search Console and to Bing Webmaster Tools; the Bing one is easy to forget and it is the index ChatGPT searches. IndexNow, a free protocol Bing supports, goes further: the site tells the engine the moment a page changes, so a new product can be findable in hours instead of waiting weeks for a crawl.

Product feeds: how shops get into shopping answers

When someone asks Google’s AI where to buy something, the products it shows come from Google’s Shopping Graph, and merchants get into it with a product feed in Google Merchant Center: a file listing each product, its price, its availability, and its photo. Meta’s shops read a similar catalog feed. If you sell physical goods, a feed that stays in sync with your site is the shortest route into those answers. It is free to submit.

Structured data: the part most sites skip

JSON-LD is a block of facts on the page written in a format machines read without guessing: this is a business, at this address, selling this product at this price, with this many reviews. Engines use it to verify what the page says. Mark up the organization, each product and service, reviews with a real rating, events, FAQs, and articles.

Answer-first pages

An engine quotes what answers the question directly. Put the answer in the first paragraph, in plain words, then explain. Write the questions buyers actually ask, in the words they use, and answer each one on its own. A FAQ that says something is worth more than a paragraph that hedges.

Consistency: the entity problem

An engine has to decide that the business on your site, the one on the map listing, and the one in a review are the same thing. Use one exact name, one description, one address and phone, everywhere. If people misspell your name, say so on the site: the sentence “sometimes spelled X” is how an engine learns the alias.

Then make the same facts agree on the profiles engines treat as reference points: your Google Business Profile if you serve customers locally, LinkedIn, and for a company Crunchbase and Wikidata. The legal name, the founding year, the town, and the one-sentence description should read the same on every one of them and on your own about page. When they disagree, the engine has to pick, and it may pick the wrong one.

Off-site signals

• Reviews on platforms the engines trust, with replies.
• Listings that agree with the site: maps, directories, marketplaces.
• Mentions in places people ask questions: forums, local press, other businesses’ pages.
• A profile on the networks your customers use, linking back to the site.

Measure it

Write five questions a buyer would ask an assistant about your kind of business without using your name. Ask them every week. Record whether the answer mentions you. That number is your share of answer, and it is the only honest way to know whether any of this is working. Questions that include your name do not count; the answer echoing your name proves nothing.

What Fraya does here

Every page Fraya publishes is server-rendered with the structured data above, written answer-first, with the business’s details kept consistent across the site. Each publish pings IndexNow, and every shop gets a Google Merchant Center feed and a Meta catalog feed, kept in sync with the site. Each week she asks AI engines real buyer questions about your business and reports the share of answers that mention you, with one-tap fixes for what would raise it. It is a measured number, not a promise. No one can guarantee a citation, and anyone who does is selling something else.

Straight answers.