There is no fixed timeline. Appearing in ChatGPT answers happens in five stages (crawled, indexed, retrieved live, cited, named), each run on someone else's schedule, so anyone who promises a date is guessing.
The first two stages depend mostly on your access settings and the crawlers' own schedules. The last two depend on what buyers ask, whether the model searches at all, and what your competitors have published, and for some questions no business gets named, ever. The rest of this page covers what each stage is, what controls it, what we have seen when we measured it, and how to avoid adding months of delay yourself.
What has to happen before ChatGPT names a business?
When people say "appearing in ChatGPT", they usually mean the last step: the assistant names their business when a buyer asks for a recommendation. Several things have to happen before that.
1. Crawled. A bot fetches your pages. OpenAI runs separate crawlers, and its documentation says the robots.txt settings for OAI-SearchBot and GPTBot are independent. You can allow OAI-SearchBot so you can appear in ChatGPT search results, and still block GPTBot from using your content for training. What controls this stage: your robots.txt, your firewall or CDN, and how many pages you offer.
2. Indexed. Your page has to sit in an index the assistant searches. OpenAI's VP of Engineering said ChatGPT Search relies on a set of services, and Bing is an important one. So a page Bing has not indexed has one fewer route into ChatGPT search. What controls this stage: Bing's crawl and index decisions, your sitemap, and whether you notify Bing of new pages.
3. Retrieved live. While answering a real person, the assistant fetches a page in real time. This is the ChatGPT-User agent. OpenAI's December 2025 documentation update clarified that ChatGPT-User fetches triggered by a user may not follow robots.txt rules. What controls this stage: whether the model decides to search at all, and whether your page comes up for the query it builds.
4. Cited. Your page shows up as a source in the answer. What controls this stage: whether your page answers the question better than the other candidates. For many questions those candidates are laws, ministries and large firms.
5. Named. The assistant mentions your business by name as a recommendation. What controls this stage: the type of question, and what other sites say about you and your competitors.
| Stage | What happens | What governs it | How you can check |
|---|---|---|---|
| Crawled | OAI-SearchBot, GPTBot and others fetch your pages | robots.txt, firewall/CDN rules, number of crawlable URLs | Server logs, filtered by user agent |
| Indexed | Page enters the search index the assistant uses, Bing in particular | Bing's crawl decisions, sitemap, IndexNow | Bing Webmaster Tools URL status |
| Retrieved live | ChatGPT-User fetches the page during a real conversation | Whether the model searches; whether your page comes up for its query | ChatGPT-User and Perplexity-User hits in logs |
| Cited | Page is listed as a source in the answer | Quality and directness of the answer versus other sources | Ask the question and read the sources |
| Named | Your business is mentioned as an option | Question type, third-party mentions, competitors' content | Repeated asks of the same buyer questions |
The stages are sequential for your own pages, but not strictly for naming. An assistant can name you because a directory or review site mentions you, even if it never read your site. Our article on how ChatGPT decides which businesses to recommend goes deeper into that last step.
What have we actually observed?
At ItGrows we put buyer questions to ChatGPT, Claude, Gemini and Perplexity every week, three times each. A few of our observations bear directly on timing.
A robots.txt mistake cost us over two months. On our own site, a rule meant to hide one redirect page (/b) was written as a prefix. Robots.txt rules match prefixes, so from 11 August to 21 October 2026 the rule also closed /blog, /business and every article in our business blog to all crawlers. Bing Webmaster Tools listed our business page as "discovered but not crawled" from 11 August 2026. Google had none of the business pages indexed. We changed the rule to /b$ on 21 October 2026. For about ten weeks the clock never started, and nothing on the page looked wrong.
Day zero was zero. On 21 October 2026 we asked 20 English and 20 German questions about getting a business into AI answers, three times each, across the four assistants. We were named in 0 of 160 answers, and none of our pages was cited. We wrote that up in our own visibility measurement.
Being indexed by Google is not the finish line. On 19 October 2026 we asked assistants for the exact titles of 18 articles on three client sites, each older than three weeks. ChatGPT found 1 of 18. Claude found 7 of 18. For one of those sites, Google had indexed 29 of 30 articles. Being in Google's index did not mean an assistant could find the page.
Size of the site matters more than markup. A site with complete markup (server rendering, Article, FAQPage and Organization JSON-LD, a sitemap) but only 75 URLs was still absent from answers on its own article's exact topic. A site with about 190,000 crawlable URLs was crawled constantly. Across the retained logs of the edtech platform pickaclass.com, 28 July to 11 August 2026, we counted 2,744,985 GPTBot requests and 19,272,125 ClaudeBot requests. A small site simply gives crawlers far fewer pages to come back for.
Live retrievals do happen. In the same fortnight the edtech platform received 1,918 live fetches from ChatGPT-User, OAI-SearchBot and Perplexity-User on content pages. A client blog of a few dozen articles received 56, spread across 24 articles. Those are moments when an assistant was reading the page while answering someone.
A first citation can come before any naming. In our weekly measurement of 20 October 2026, Perplexity cited an article we had published on a client's site, a German company-formation advisory. That is stage four. It does not yet mean the firm gets recommended.
Some clients do report business results. FreightSizer, a B2B SaaS for freight dimensioning, reported 2 new B2B customers in one month, both from ChatGPT. MagiScan, an iOS 3D-scanning app, reports hundreds of installs a month from ChatGPT, Claude and Gemini. appslift, an app store optimisation service, reports 2 to 5 leads a month from AI assistants. These figures are reported by the clients themselves, and they show that the end of the road exists, not how long the road is.
Why can't anyone promise a date?
Every stage depends on a decision you do not control.
- Crawl schedules belong to the crawlers. Bing decides when to move a URL from "discovered" to "crawled". Our business page sat in that state for weeks, though in our case we had caused it.
- The model decides whether to search. In a test of 20 German business questions on 23 September 2026, the model answered 10 without searching at all and cited nobody. If it does not search, your fresh page cannot appear, however good it is.
- Primary sources win informational questions. In the same test, where the model searched, it cited primary sources (laws, ministries, OECD, large accounting firms) on 9 of 10 questions. Firms were named only on "who should I hire" or "how to choose" questions.
- Answers vary between asks. Two runs of the same question named two different sets of firms. Being named once is not a stable position.
- Competitors are publishing too. For an orthopaedic practice in Bonn, measured on 20 October 2026, the assistants named the practice in 12 of 80 answers. Competing practices' own websites were the most cited sources, and the directory jameda.de was cited in 12 answers. Questions about costs, risks and treatments named no practice at all.
So an honest answer is a sequence of conditions, not a date. If someone guarantees you will be recommended by a certain week, ask them which of these stages they control.
What can you do so you are not the bottleneck?
You cannot speed up Bing or OpenAI, but you can make sure none of the delay is self-inflicted.
- Open the door to the right bots. Allow OAI-SearchBot at minimum. Perplexity's documentation recommends allowing PerplexityBot in robots.txt and permitting its published IP ranges. Check for prefix rules like ours, and check your CDN or firewall, which can block bots even when robots.txt allows them. Our guide on whether your site is blocking GPTBot, ClaudeBot and PerplexityBot walks through this.
- Get into Bing. Verify the site in Bing Webmaster Tools and submit your sitemap. Microsoft's IndexNow page describes four steps: generate an API key, host the key file, submit URLs, and verify in Bing Webmaster Tools. That way Bing hears about new pages without waiting for its next crawl. More on this in Does ChatGPT Use Bing?.
- Publish pages that answer buyer questions. Naming happens on "who should I choose" questions. Write pages that answer them directly, with specifics a buyer can compare: who you serve, what it costs, how you differ.
- Grow the surface. A handful of perfect pages is less visible than a steady body of useful ones. Each article is another chance to be crawled, retrieved and cited.
- Be present where assistants already look. In the Bonn example, a directory was among the cited sources alongside competing practices' websites. If assistants keep citing a particular directory or review site in your field, make sure your listing there is accurate and complete.
How will you know when it has started?
Do not wait for website traffic as your signal. In about two weeks of logs from three of our course websites, roughly 1.2 million requests in the two busiest logs, human visitors arriving from AI assistants totalled one, from Copilot. Assistants mostly answer in place rather than send a click.
Measure the stages instead:
- Crawled and retrieved: look for OAI-SearchBot, GPTBot and ChatGPT-User in your server logs.
- Indexed: check URL status in Bing Webmaster Tools.
- Cited and named: ask a fixed set of buyer questions, several times each, and record who is named and which sites are cited. We count a business as named if any of three asks names it. We compare weeks only on the same questions with the same number of asks. With October 2026 API prices, a report of 20 questions across 4 assistants with 3 asks each costs us about $5.40.
Our guide on how to check whether AI assistants mention your business shows how to set this up yourself. If you want a quick baseline first, the free check at itgrows.ai generates five buyer questions from your homepage and shows in about 30 seconds whether ChatGPT and Perplexity name you. ItGrows also does the full work for $499 a month (plus a one-time $499 setup, free with 3 months or a year): publishing articles on your domain, fixing crawler access and indexing, and measuring the four assistants weekly. Nobody can guarantee that an assistant will name a business, and we do not.
Frequently Asked Questions
Can I appear in ChatGPT answers within a few days?
For a new page on a small site, we would not count on it. In our 19 October 2026 check, ChatGPT found only 1 of 18 client articles older than three weeks, even though Google had indexed 29 of 30 articles on one of those sites. Each stage has to happen first, and you control only part of it.
Does being on Google's first page mean ChatGPT will find me?
Not necessarily. ChatGPT search relies on several services, with Bing an important one, so Bing indexing matters on its own. We have seen pages indexed by Google that assistants still could not find by their exact title.
Why does ChatGPT name me one day and not the next?
Answers vary from ask to ask, and two runs of the same question can name different businesses. That is why we ask each question three times and compare weeks only on the same questions. A single screenshot tells you little.
What is the most common reason the clock never starts?
Blocked access is the one to rule out first. A robots.txt rule, a CDN bot setting or a firewall can stop crawlers without anything looking wrong on the page. Our own prefix rule closed our blog to all crawlers from 11 August to 21 October 2026, so check your access before anything else.