To get cited by ChatGPT and Perplexity, start with access, then publish what only you can.
Six moves, ranked by how much evidence supports each. Some are documented by the platforms, and others rest on studies that show correlation. I label which is which.
To get cited by ChatGPT and Perplexity, let both crawlers in first, then publish something only you can supply.
After that, answer your buyer’s follow-up questions inside that page, get named on sites your buyers already read, keep your facts current, and check the same buyer questions every month.
That is the order I would work in, and so far it’s done pretty well for my clients — both the larger enterprise companies I work with and small businesses who feel like they can’t break through the noise very well.
The key is to grade each move by how much evidence sits behind it.
Most advice on this topic lists twenty tactics and grades none of them. You finish the article unable to tell what the platforms document from what a vendor sells.
If you read the first post in this series, you know why a page that ranks on Google can still go uncited. Now we’ll cover what to do about it, in what order, and what you can safely ignore.
Remember, nobody can promise you a citation, and I mark the point where each move stops being proven.
- Start with access. OpenAI says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, and Perplexity asks you to allow PerplexityBot.
- Publish what other sites summarize. Ahrefs’ October 2026 data suggests ChatGPT is moving toward the sources that produce information.
- Freshness helps, with limits. AI-cited pages averaged 1,064 days since publication, so the typical cited page is nearly three years old.
- Nobody can promise a citation. Track the same buyer questions every month and judge the moves by your own data.
What are the six moves, in order?
Here are the six moves in order. Each one carries one of three grades, and I’ve found that these cover things pretty well across the board.
Documented means the platform itself says so.
Supported means a large study found a pattern, usually a correlation and often from one model in one month.
Position means my judgment, backed by whatever I have observed myself.
| Move | Evidence | Source |
|---|---|---|
| 1. Let OAI-SearchBot and PerplexityBot in | Documented | OpenAI, Perplexity |
| 2. Publish what other sites summarize | Supported | Ahrefs, Oct 2026, Google |
| 3. Answer the follow-up questions inside one strong page | Supported, with a warning | Ahrefs, Oct 2026, Google |
| 4. Get named on sites your buyers read | Supported, correlation only | Ahrefs, 75K brands |
| 5. Update when facts change, and show the real date | Supported | Ahrefs, 17M citations |
| 6. Track the same questions every month | Documented for Google, my position for the rest |
Only the first move has an “on and off switch”. The other five are judgment calls with different amounts of evidence behind them.
Can ChatGPT and Perplexity reach my site?
Only if you let their search crawlers in. This is one of the most important things to ask. And nobody thinks to check on this.
OpenAI says sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, although they can still appear as navigational links.
Perplexity says PerplexityBot is designed to surface and link websites in search results, and it asks you to allow the bot in robots.txt.
Perplexity also asks you to permit its published IP ranges, which is why I would check your CDN and firewall rules as well as robots.txt.
OpenAI says a robots.txt change can take about 24 hours to take effect, so make the change and look again the next day. One more detail from Perplexity: Perplexity-User, which runs when a person asks a question, generally ignores robots.txt.
Access puts your pages in the pool an engine can cite. It does not make an engine choose you.
Evidence grade: Documented.
What should I publish that an engine would cite?
Publish the thing other sites summarize. Original data, a named method, or a first-hand case gives an engine a reason to cite you and not the ten pages repeating you. Basically, what makes YOU YOU — and what makes it worth doing business or partnering with you.
We live in a world where anyone can say anything with AI. In fact, I used AI to search up most of these stats! (and double-check them, of course.)
Ahrefs’ October 2026 study of 66.6 million ChatGPT responses found that ChatGPT “seems to be moving from sites that summarize information to the sources that produce it.”
Here is what moved:
PubMed rose from 64th to 4th among the most-cited domains
Reddit’s share fell about 88%, from roughly 1 in 5 answers to 1 in 43
Wikipedia’s fell 80%
On about 25,000 identical prompts present in both months, every site in Ahrefs’ chart moved in the same direction.
The limits matter here too: one model, a free logged-out setting, and two months of data. Ahrefs says “seems,” so I say “suggests.”
Give your own facts an official page too: what you do, who it is for, who it is not for, and how you price. In the same study, 84.4% of fact questions cited at least one official source, against 68.3% of comparison questions.
The brand-site figures there are led by Apple, IKEA, and Nike, so a small B2B company should not expect those numbers.
I have said that AI cites the most recognizable source, and this data suggests who counts as recognizable: the origin of the information.
That is the first stage of the Citation Authority Flywheel —>publish what only you can. My receipt is one HOA software client. In a sample of 10,709 Google AI Mode answers, about 39% cited the client, and it was the most-cited source in all 10 core topics.
That is one observed case in Google AI Mode, not ChatGPT or Perplexity, so treat it as a data point and not a formula. For what to put on the page itself, see what to include in content to get cited by AI.
Evidence grade: Supported.
If ChatGPT is moving toward the sources that produce information,
the job is to produce something only you can say.
— Brad Bartlett
Do I need a page for every question my buyers ask?
No. Cover the follow-up questions inside one strong page.
Think about it: engines can run extra searches behind the scenes before they answer, and Ahrefs measured that in 14.4% of ChatGPT responses, averaging about 1.6 extra queries each.
That makes it a refinement to make after the first two moves, not a first step.
Definition
Query fan-out is “a set of concurrent, related queries generated by the model to request more information.”
Google’s definition, from its generative AI optimization guide
Google’s guide warns against creating pages for query variations “primarily to manipulate rankings or generative AI responses.”
So answer the obvious next questions on the page you already have. Ahrefs also found that “2026” appeared in 17.6% of fan-out queries in August and 22.9% in September, and “official” rose from 10.1% to 13.1%.
That suggests engines look for current, official sources, though the prompt mix is broad consumer demand and not B2B.
The mechanics of writing each answer are in how to write content that gets cited by AI systems.
Evidence grade: Supported, with a warning.
Does getting mentioned on other sites help?
Plausibly yes. But I’ve found that the evidence is more correlation than causation.
Across 75,000 brands, Ahrefs found that branded web mentions correlated with Google AI Overview visibility at 0.664, against 0.218 for backlinks.
The study covers AI Overviews only, and it says plainly that correlation is not causation. It says nothing directly about ChatGPT or Perplexity.
The engines also disagree about where to look. In Ahrefs’ June 2025 data, only 7 of the top 50 most-mentioned sources were shared across Google AI Overviews, ChatGPT, and Perplexity.
| Google AI Overviews | ChatGPT | Perplexity | |
|---|---|---|---|
| Where it leans for sources | Top search sites, plus health, finance, encyclopedic, and social sites, and Google-owned properties | Publishers and non-Google media, partly through licensing deals | A broader international set, including local and regional brands |
| Age of cited pages vs. organic Google | 16 days older | 393 to 458 days newer | 250 days newer |
| Crawler that gates visibility | Googlebot, and Google says no special files are needed | OAI-SearchBot | PerplexityBot |
Source lean: Ahrefs, June 2025. Age of cited pages, by publication date: Ahrefs, July 2025. Both are 2025 data from one tool each.
So match the placement to the engine your buyers use.
Do not count on Reddit alone: its share of ChatGPT answers fell about 88% in the Ahrefs dataset above.
The best receipt I have for off-site mentions is human. A firm with 100,000 units emailed asking to be added to an article. A person read the mention and decided it was worth having, which is the same judgment an engine is trying to approximate.
The how-to for earning mentions is in what brand signals get you cited in AI answers.
Evidence grade: Supported, correlation only.
Do I need to keep updating content to stay cited?
Update when your facts change, and show the real date.
Freshness is a preference and not a requirement. Across 16.975 million cited URLs, AI-cited pages averaged 1,064 days since publication, against 1,432 for organic Google results.
The typical page an AI assistant cites is nearly three years old. (Which, yes, is crazy.)
days since publication, on average, for a URL an AI assistant cited
Organic Google results averaged 1,432 days. Ahrefs counted 16.975 million cited URLs. Source: Ahrefs, July 2025.
By last-updated date the gap is smaller: 909 days against 1,047.
By platform, ChatGPT’s in-text references ran 393 days newer than organic Google results, Perplexity’s ran 250 days newer, and Google AI Overviews ran 16 days older.
Treat those as 2025 data from one tool. Publication age is a crawler’s first-seen date, and the Ahrefs study’s author suspects most brands gain more from new, high-quality content than from frequent updates.
The study also notes that Google’s John Mueller warns against changing dates without changing the content. Do not stamp a new date on an old post. My advice on structuring a page for recency is in how to structure a page for AI citations.
Evidence grade: Supported.
How will I know it’s working?
Pick 10 to 15 buyer questions, ask them in each engine in the same words, and log who gets cited.
It takes work, yes, but do it. Then repeat every 30 days.
For the Google side, Search Console’s Generative AI report covers what Google shows. Nobody gives you a dashboard for ChatGPT and Perplexity yet, so the log is the measurement.
I run this on myself. For instance, the question this post answers, “How do I get my B2B company’s content cited as a source in ChatGPT and Perplexity answers instead of just ranking on Google?”, returned no citation for anyone as of October 28, this site included.
My own citation log has no rows yet, and I will check the question again after this publishes. Whether a citation sends traffic is a separate question, covered in does AI citation drive traffic.
Evidence grade: Documented for Google, my position for the rest.
What should I skip?
Skip llms.txt, special AI markup, and cutting pages into tiny chunks, at least for Google.
Google’s guide says “You don’t need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search.”
It also says there is no requirement to break your content into tiny pieces, and that structured data isn’t required for generative AI search.
Skip it or do it
Same effort. Different evidence.
Tactic
llms.txt, special AI markup, and cutting pages into tiny chunks
What the source says
Google says none of it is needed to appear in Google Search
The gap
I found no statement from OpenAI or Perplexity either way
Tactic
Crawler access, original content, honest dates, monthly tracking
What the source says
Platform documentation, Ahrefs studies, and Google’s guide
The gap
Several rest on correlation, so test them on your own questions
The sources behind both columns:
I wrote “Google says” on purpose here. I found no equivalent statement from OpenAI or Perplexity, so the honest label for those tactics is unproven, not useless.
Skip self-promotional “best” lists as a tactic, too. Ahrefs found them in 34% of software prompts where the publisher was recommended, with no downside measured, but that is co-occurrence and not proof they helped.
Seer Interactive saw listicles slip from 17.8% to 15.5% of ChatGPT citations between November and January. Seer’s 30% headline is a one-month drop in count, and Seer says so.
If you want to know which formats do get cited, start with what content formats get cited by AI.
What would I do in the next 30 days?
This is my judgment, and the hours are my estimates. Adjust them for your team.
Week 1
Open the door and set a baseline
Check robots.txt and your firewall for both crawlers (about 1 hour). Run 10 to 15 buyer questions in ChatGPT, Perplexity, and Google, and log who gets cited (2 to 3 hours). Write the page that states what you do, who it is for, and who it is not for (about 3 hours).
Weeks 2 and 3
Publish one original asset
A dataset, a benchmark, or a documented client case, anonymized if you need to. Answer the next three questions a buyer would ask inside it.
Week 4
Earn three placements
Three mentions on sites your buyers already read. Match each placement to the engine you want to reach.
Day 30
Re-run the same questions
Same wording, same engines. Log what changed and what did not, then decide what to repeat.
That is one afternoon of access and baseline work, one real asset, three placements, and a re-run. If the numbers move, repeat what moved them. If they do not, you will know which move to question.
Which engine should I optimize for first?
The one your buyers use. Ask your last five customers which AI tools they used while researching you, and start there.
You must know where your people are most likely to search you or about you. And that requires having a clear understanding of your market fit and Ideal Client Personas.
Not sure where you stand? Answer five yes-or-no questions below.
Is your site set up to be cited?
Five honest questions. Your score reveals a verdict at the bottom, and no email is required.
1. Have you checked that OAI-SearchBot and PerplexityBot are not blocked on your site?
2. Does one page on your site state what you do, who it is for, and who it is not for?
3. Have you published original data, a benchmark, or a documented case in the last 12 months?
4. Is your company named on at least three sites your buyers read that you do not control?
5. Do you check the same 10 or more buyer questions in AI engines every month?
Can anyone promise this will get me cited?
No, and I’d be wary of anyone who does. Most are trying to sell you something, and this is still a bit of a black box. The hacks and schemes people peddle today could fall apart overnight.
The fine print
Nobody can promise a citation, and that includes me. These studies measure correlation, one model at one moment, or a month of data, and the engines change how they retrieve sources often. Treat each move as something to test against your own buyer questions.
That applies to this post too. Your results depend on your buyers’ questions, your category, and what the engines change next month.
Give me the short version, Brad.
Let the crawlers in. Allow OAI-SearchBot and PerplexityBot, and check your CDN and firewall.
Publish what only you can. Original data, a named method, or a first-hand case, plus an official page of your own facts.
Answer the follow-ups on one page. Do not spin a page per variation.
Get named where your buyers read. The evidence is correlation, so match placements to the engine.
Keep facts current and dates honest. New, good content likely beats cosmetic refreshes.
Track the same questions every month. The log is your measurement.
Sound human. Get cited.
If your buyer asks ChatGPT who to hire, is your name in the answer?
I check three things: whether ChatGPT and Perplexity can reach your site, what they cite for your buyers’ questions instead of you, and which of these six moves should come first for your business.
Start with a GEO Visibility Audit. You get a list of what to fix first, in order.
Book a GEO Visibility Audit →Fiverr Pro vetted · 4.9 stars · 1,600+ client reviews
Made with 💙 in kcmo
Frequently Asked Questions
How long does it take to get cited by ChatGPT?
No source I found gives a timeline. OpenAI says a robots.txt change can take about 24 hours to take effect for search, but that only decides whether you can appear. Whether you get cited depends on your content, so I would judge it on a monthly check over at least a quarter.
Do I need llms.txt to get cited?
Not for Google. Google’s guide says you don’t need new machine readable files, AI text files, markup, or Markdown to appear in Google Search. I found no equivalent statement from OpenAI or Perplexity, so I would not spend time on it first.
Does Perplexity cite different sources than ChatGPT?
Yes. In Ahrefs’ June 2025 data, only 7 of the top 50 most-mentioned sources were shared across Google AI Overviews, ChatGPT, and Perplexity. ChatGPT leaned on publishers and non-Google media, and Perplexity pulled from a broader international set.
Is ranking on Google still worth it?
Yes, for Google search itself. Nothing I found shows that rankings carry over to ChatGPT or Perplexity, and in Ahrefs’ June 2025 data only 7 of the top 50 most-mentioned sources were shared across Google AI Overviews, ChatGPT, and Perplexity. The first post in this series explains why a page that ranks can still go uncited.
Does schema help me get cited?
Google says structured data isn’t required for generative AI search, and that you don’t need new markup to appear in Google Search. I found no equivalent statement from OpenAI or Perplexity, so I would not treat schema as a citation tactic.
Should I block AI crawlers?
If you want to appear in ChatGPT search, do not block OAI-SearchBot. OpenAI says sites opted out of it will not be shown in ChatGPT search answers. OpenAI documents GPTBot separately, for model training, so allowing search and declining training are two different choices.
How do I check whether ChatGPT cites my site?
Ask it your buyers’ questions in the same words each time, and log whether your brand or URL appears. Repeat monthly. For Google, the Generative AI report in Search Console covers that side.
