If you are asking how to rank in ChatGPT, the honest answer from Benian Technologies is that there is no fixed rank to win: you raise the odds that ChatGPT and Google AI Overviews name your business by letting their crawlers reach your pages, publishing one consistent set of facts, marking those facts up clearly, answering the questions buyers actually ask, and being mentioned on sources those systems read. None of those steps forces a citation. Together they remove the reasons an answer engine has to skip you.
The cost of skipping this work is quiet. A buyer asks an assistant which firm, shop or supplier to call, gets three names, and never visits a search results page where you might have appeared. You do not see the lost lead in any report. Below is the action list in the order we would do it, including the claims you should refuse to pay for.
Benian is an AI implementation partner, and AI Visibility is one of eight services it sells. The free scan at /tools/ai-visibility checks twelve technical signals on your site with no signup. Fix work after that is scoped from the findings, and Benian publishes no price for it or for anything else.
How to rank in ChatGPT when there is no fixed ranking
A Google results page is a list with positions. A ChatGPT answer is generated text. When it searches the web, it pulls a handful of pages, reads them and writes a reply that may cite some of them. Two people asking the same question on the same day can get different names. A model update can change which sources it prefers. So any report that shows you at position one in ChatGPT is describing one sampled answer, not a stable rank.
Google AI Overviews sit closer to classic search. Google has said a page needs to be indexed and eligible to show a snippet in normal results to be used as a supporting link, and that no special markup is required. That is why seo for ai search is mostly ordinary search work done carefully, plus attention to how clearly a page states facts that a model can lift and attribute.
What you can change is the inputs: whether the systems can fetch your pages, whether what they find is clear and consistent, and whether other sites confirm it. That is the whole of a sound aeo strategy. Anything sold as a shortcut past those inputs is selling a sample, not a result.
Step one: let AI crawlers reach your pages
Open your robots.txt file and read it line by line. OpenAI documents separate crawlers: GPTBot collects training data, while OAI-SearchBot fetches pages for ChatGPT search results. Blocking GPTBot keeps your content out of training. Blocking OAI-SearchBot can keep you out of ChatGPT search answers. Many sites block both by accident, through a broad rule copied from a template or a firewall setting that challenges unfamiliar bots.
For Google, AI Overviews draw on the normal Googlebot index. The Google-Extended token controls use of your content for Gemini training, not whether you appear in AI Overviews. Decide each crawler on purpose and write the decision down.
Then check what the crawler actually receives. If your prices, service areas or product details only appear after JavaScript runs, some crawlers will see an empty shell. Fetch the raw HTML of your key pages and confirm the facts are in it.
Step two: one set of business facts, everywhere
Answer engines cross check. If your site says you serve three states, your Google Business Profile says two and an old directory listing gives a different phone number, the model has to pick one or leave you out. Write a single fact sheet: legal and trading name, address or service area, phone, hours, what you sell, who it is for, shipping or delivery terms and the policies a buyer asks about. Then make the website, your Google Business Profile, your social profiles and the main directories in your industry match it exactly.
This is the same discipline any assistant needs. VOT Distribution, a multi-brand e-commerce distributor, runs two AI storefront assistants built by Benian, including one at shopfreezo.com that answers product, compliance and shipping questions. An assistant like that can only be as accurate as the product and shipping facts it is given, and a public answer engine reading your site is no different. The linked VOT case study covers that engagement and its labeled figures; it is not an AI visibility result, and we do not present it as one.
Assign an owner. Facts drift when a delivery window or service area changes and only one of five places gets updated. A quarterly check against every listing prevents most of it.
Step three: structured data and pages that answer buyer questions
Structured data is code on the page, usually JSON-LD using the schema.org vocabulary, that labels facts: this is the organization, this is the product, this is its price, this is the question and this is the answer. It does not make a model cite you. It removes ambiguity about what your page says, and it must match the visible text or it does more harm than good. Start with Organization or LocalBusiness on the home page, Product on product pages and Article on guides.
Then write for the question, not the keyword. Collect the questions buyers ask your sales team, support inbox and phone line. Give each important question its own page or section, put the direct answer in the first two sentences, then the detail: steps, exceptions, costs and when the reader should choose something else. A model quotes a sentence that stands on its own, not one that needs three screens of brand story first.
This is where ai overviews seo and good sales writing converge. A page that names its own limits and the cases where the buyer should not buy is easier to trust and easier to quote than one that claims to suit everyone.
Step four: mentions on sources AI systems read
Models lean on what other sites say about you. Useful sources are the ones a buyer in your industry would trust anyway: trade associations, industry directories, review platforms, local news, partner and supplier pages, and forums where your buyers ask questions. A mention that states what you do and where you do it is worth more than a bare link.
Earn these the slow way: answer questions in public, publish original data from your own operation, and ask customers and partners to describe your work accurately. Avoid paid link schemes and mass directory submissions. They tend to create inconsistent facts, which works against step two, and search engines treat many of them as spam.
What an AI visibility score measures, and a first month checklist
An ai visibility score is only as meaningful as what it counts. Benian's free scan scores twelve technical signals in the page HTML and supporting files your site returns, such as robots.txt and structured data, and lists findings and next steps. It does not measure rankings, citations or recommendations. Tools that claim to score your presence inside ChatGPT are sampling answers to a set of prompts, so ask which prompts, how often and how the samples are recorded before trusting the number.
Optional monitoring can report requests matching crawler user-agent labels. Treat that carefully: a label can be faked, and a visit does not prove indexing or a citation. For citations, run a fixed list of twenty or so buyer questions on a fixed schedule, save every answer in full, and read the trend over months.
A first month: week one, audit robots.txt, firewall bot settings and raw HTML. Week two, write the fact sheet and correct every listing. Week three, add structured data that matches visible text. Week four, rewrite the five pages buyers need most so each opens with a direct answer, and start the question log. When not to hire anyone: if your site is small and your facts are already consistent, this list is work you can do yourself. Bring in help when the fixes touch a site build, a firewall, or hundreds of product pages.