ChatGPT cites a page when it searches the web for a question and finds a page that answers it plainly near the top. That is how to get cited by ChatGPT: publish that page for each question your buyers ask, let AI crawlers in, and ask the question again after it goes live.
Your buyers have started asking an assistant which product to use, and the answer names three companies. If yours is not one of them, you never see the visit, the search impression or the bounce. Nothing shows up in your analytics at all. That is why getting cited by ChatGPT has become a real budget line and not a curiosity, and why most advice about it is so unsatisfying: it is written as a checklist of tactics with no way to tell whether any of them worked. This piece covers the mechanism, the five things that move citations, and how to tell a change you caused from drift you did not.
How to get cited by ChatGPT, and how to know it worked
An assistant's answer is not a ranking. Ask the same question twice and you can get two different sets of names. Ask it from a different account, or a week later, and it moves again. There is no rank tracker that makes this stable, because the instability is in the thing itself, not in your instrument.
That has two consequences, and both are practical.
The first: a single check tells you almost nothing. One answer that names you is one draw from a spread of answers you have not seen. You need the same question asked many times before "we get mentioned" becomes a number rather than an anecdote.
The second: a change over time is not evidence of your work. These platforms update on their own schedule. Something you did in week one and something they shipped in week three both show up as the same movement in the same chart. Untangling them is the entire job. On Scale and Agency, Rankli does it by comparing the questions it worked on with ones it left alone: a platform change moves both groups, and your work shows up only in the first.
Does AI name your business?
Ask ChatGPT and Perplexity three questions your buyers ask. Free, no signup.
Where a citation actually comes from
Two paths, and they respond to different work.
From the web
The assistant runs a search at answer time, reads a handful of pages and writes from them. Being cited here means being one of the pages it reads, and being useful enough in that moment to survive into the summary. This path moves fast, and ordinary publishing can reach it. It also rewards a specific shape of page: one that answers a named question directly, near the top, in language that matches how the question was asked.
The entry condition is written down. OpenAI's crawler documentation, read on 29 September 2026, says: "Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links."
From the AI's memory
The model answers from what it absorbed before the conversation started. Being named here is a slower business, and it is downstream of how often your product is described on pages that other people wrote: comparison posts, directories, forums, roundups, documentation that quotes you. You cannot edit that corpus, but you can be worth including in it, and you can make the facts about you consistent everywhere they appear.
An answer can mix the two. The practical reading: publishing reaches the web half first, and the memory half follows the reputation the web half builds.
The five things that move AI citations
Answer the exact question, high up the page
Assistants summarize. A page that spends 400 words warming up gives the summarizer nothing to lift. Put the direct answer in the first paragraph under the heading that matches the question, then earn the rest of the read. This is also the structure our scorer enforces: the target phrase has to appear in the title, in the first 100 words and in at least one H2, which is a crude proxy for exactly this shape.
Be present where the comparisons happen
When someone asks an assistant for the best tool in your category, the pages that answer are roundups, alternatives lists and comparison posts, and other people write many of them. Being nameable there is a content job: publish the comparison page for your own category, keep every competitor figure sourced and dated, and be the page that is worth citing rather than the page that begs to be. Our own comparison hub is built that way, with each figure read off the vendor's live page on a stated date.
Let AI visit your site
Nothing gets cited that cannot be fetched. Check the settings that let AI visit your site, in your robots.txt file, for the assistants' crawlers by name. Check that your key pages arrive with their text in them rather than being assembled in the browser. Check that nothing important sits behind a pop-up. Rankli's free check runs all three on your site and asks no AI assistant anything. Rankli's own crawler behavior is documented on our bot page, and the same discipline applies to whoever is fetching you.
Keep your facts identical everywhere
If your pricing page says $99 and your homepage implies $79, a model that read both can state either one. Pick the canonical number, put it in one place in code if you can, and render it everywhere from there. Consistency keeps a machine from repeating a version of you that you have already retired.
Publish at a cadence, on the questions buyers actually ask
How often AI cites you follows coverage: you get named on questions you have written something useful about. Thirty articles a month on the questions your buyers ask is a different asset from thirty articles a month on whatever had search volume. This is why Rankli plans part of every month at the questions you are losing rather than from a generic list, and why every draft has to mention what you actually sell to pass its checks.
How to measure it instead of guessing
Pick the questions, then split them
Start from the commercial questions a buyer would type: "best X for Y", "X vs Z", "is X worth it". Rankli drafts them from your site, your category and your competitors. On Scale and Agency, it then splits them into two groups before any measuring happens: questions it will work on and questions it will leave alone. The split has to come first, because a comparison group chosen after you have seen the results proves nothing.
Ask each question every week, on every engine
A single answer is noise. Rankli asks each tracked question every week on ChatGPT, Claude and Gemini, and on Perplexity too on Scale and Agency. It then reads each answer for which brands appear and which pages it cites. On Growth that is 50 tracked questions a week on three engines, and on Scale 150 on four, which is enough to put an error bar on the number rather than report a single reading.
Leave some questions alone, and write the prediction down first
On Scale and Agency, before the work starts, Rankli writes down the change it will make, the questions it should move, the size of the effect it expects and the questions deliberately left alone. Then it asks again and compares the two groups. The readout is the change in the questions it worked on minus the change in the ones it left alone, which is where platform drift cancels out.
Ask again after it goes live
On every plan, 7 and 14 days after an article goes live, Rankli asks the article's question again, 3 times on each engine your plan tracks. The article then shows before and after: how many answers named you before it went live and after, and whether AI cited the page.
Read the difference, not the line
On Scale and Agency, both numbers are shown, and they can disagree. That is the point. A before-and-after line moves when the platform moves, whether or not you did anything. The difference between the questions you worked on and the ones you left alone moves when your work did. When that difference is too small to separate from noise, Rankli calls the result inconclusive rather than rounding it into a win.
What quotable pages have in common
The pages an assistant can quote share a family resemblance, and it is not a subtle one.
They are specific. A page called "Outrank alternatives" with nine products, each with a price read off a live vendor page and a date attached, gives a summarizer something to quote. A page called "the future of content marketing" gives it nothing.
They are structured for extraction. One question per heading, the answer directly underneath, lists where a list is the shape of the answer. Assistants lift paragraphs, not vibes.
They carry numbers that can be checked. Prices, limits, counts, dates. A page that says "affordable" competes with everyone; a page that says "$99 a month for 30 articles, read on 4 September 2026" is the one worth citing.
They state who is speaking and when. A visible publication date, a named organization, and figures that agree with the rest of your site. This is also what our own scorer tests on every draft, from the other direction: an article with an unsourced statistic in it fails a hard rule, which means it never becomes one of these pages in the first place.
None of that requires writing for machines. It is what a well-made reference page has always looked like. The change is that a badly made one now loses in a place you cannot see.
A four-week routine that fits in a normal week
Week one: fix what stops AI reading your site, and make your facts consistent. One canonical price, one canonical product description, AI crawlers allowed, key pages served with their text in them.
Week two: publish the two pages that answer your category's most-asked buying questions directly, with the answer in the first paragraph.
Week three: publish your comparison pages, sourced and dated, and correct the facts about you on any third-party listing you control.
Week four: ask your questions again and compare the answers with the ones you recorded before you started. Be willing to conclude that a change did nothing.
Then repeat. How often AI cites you moves slowly, in a market where every competitor is also publishing, and the compounding comes from doing this every month rather than from any single clever move.
What to avoid
Do not buy a tool that reports a single citation score with no range and no comparison with questions it left alone. It will move on its own and you will make decisions on the movement. Do not write pages aimed at models rather than people, because the summarizer is reading the same words your buyer is. And do not treat one flattering answer as a result: ask the same question ten times before you believe it.
Rankli asks ChatGPT, Claude and Gemini your buyers' questions on every plan, and adds Perplexity on Scale and Agency. Every plan starts with a 7-day free trial, and starting it runs your first analysis. On day 8 it becomes your plan unless you cancel, which is one click in the app, and cancelling during the trial means nothing is charged. Plans and limits are on the pricing page.
Does AI name your business?
Ask ChatGPT and Perplexity three questions your buyers ask. Free, no signup.
Questions
How long does it take to get cited by ChatGPT?
There is no fixed time: it depends on when ChatGPT next reads your page and on what else it finds. Rankli asks each article's question again 7 and 14 days after it goes live, so you see whether the page changed the answer.
Can ChatGPT cite a site that blocks its crawler?
OpenAI's documentation says sites opted out of OAI-SearchBot are not shown in ChatGPT search answers, though they can still appear as navigational links. Rankli's free check reads your settings for OpenAI's crawlers by name.
Is getting cited the same as being recommended?
No. A citation links your page as a source, and a recommendation names your business in the answer. Rankli records both for every answer, so you see which one moved.
How does Rankli know a change came from my article?
On Scale and Agency, Rankli compares the questions it worked on with ones it left alone, so a platform change that moves both is not credited to your work. On every plan, it asks each article's question again 7 and 14 days after it goes live and shows before and after.