Sauce · AI search
How to Get Cited by ChatGPT: A Step by Step Guide
Quick answer
To get cited by ChatGPT, let OpenAI’s search crawler read your site, then give it a page worth citing. Allow OAI-SearchBot in robots.txt. Make each page easy to crawl. Put the answer in the first two lines. Show who you are and where you work. Keep your profiles the same everywhere. Then ask ChatGPT real questions and write down what it cites. OpenAI says placement is not guaranteed. We say the same.
Can ChatGPT read your site today?We take one client per trade, per city. The call starts with that check, not a pitch.
Book a free assessmentChatGPT can only cite a page it can read.
Picture a homeowner in Surrey. A storm just peeled shingles off her roof. She opens ChatGPT and types, “Who can fix a roof in Surrey this week?” She gets a short list of names, each with a link.
Your name is not on it. Maybe your site blocks the crawler that feeds that list. Maybe the answer is buried on page four. Either way, she never sees you, and you never see her.
This guide is the fix, in order. It has ten steps, a plain test for each, and a 14-point checklist at the end. It is written for owners of local service businesses. You do not need to code. You do need to be able to ask your web person one clear question.
How does ChatGPT choose which websites to cite?
OPENAI DOCUMENTEDChatGPT search turns your question into one or more targeted searches. It may send follow-up searches too. It then reads what comes back and links the sources it used. OpenAI adds that it ranks results using several factors, and that placement is not guaranteed. It also warns that citations can be incomplete, outdated or wrong.
No one outside OpenAI knows the full ranking recipe. We do know the order of events: your page has to make it through each stage.
You controlChatGPT decides
-
Crawl
OpenAI’s search crawler visits your page, if your site lets it in.
You control -
Match
ChatGPT turns the question into searches and finds pages that fit.
You control the fit -
Rank
It weighs several factors. OpenAI has not published the list.
ChatGPT decides -
Cite
It links the sources it used. Placement is not guaranteed.
ChatGPT decides
Steps 1 to 10 below work on the first two stages and on the proof that helps with the third. We cannot touch the fourth. Nobody can.
Let OpenAI’s search crawler in
OPENAI DOCUMENTEDOAI-SearchBot is the crawler that finds sites to show in ChatGPT search answers. Sites that opt out will not be shown in those answers, though they can still appear as plain links. OpenAI recommends allowing OAI-SearchBot in robots.txt.
Here is the rule. Add it to the robots.txt file at the root of your site.
robots.txt
User-agent: OAI-SearchBot
Allow: /
Then run three checks.
- Open the file. Visit
yourwebsite.com/robots.txt. Look for a line that saysDisallow: /underUser-agent: *or under OAI-SearchBot. A site built on a staging copy often keeps that line by mistake. - Check your host and CDN. OpenAI says to allow its published IP ranges, which are listed at openai.com/searchbot.json. A firewall or bot shield can block the crawler even when robots.txt allows it. Send that link to your web person.
- Wait about a day. OpenAI says its systems may take about 24 hours to adjust after a robots.txt update.
OpenAI runs more than one crawler. They do different jobs, and each setting is independent. Blocking one does not block the others.
| Crawler | What it does | Sets ChatGPT search? | Our suggestion |
|---|---|---|---|
| OAI-SearchBot | Finds sites to show in ChatGPT search answers. | Yes | Allow it. |
| GPTBot | Collects content that may be used to train OpenAI’s models. | No | Your choice. It does not decide search. |
| ChatGPT-User | Visits a page when a person asks ChatGPT to. OpenAI says robots.txt rules may not apply. | No | Know it exists. Do not rely on robots.txt for it. |
We checked our own site
DHM DATAOn November 19, 2026, our live robots.txt allowed OAI-SearchBot, GPTBot and ChatGPT-User by name, along with nine other AI crawlers. We keep an llms.txt file as well. Neither Google nor OpenAI lists it as a requirement in the pages we read, so it is not why we get found.
Make every page easy to crawl
A crawler that is let in still needs something to read. Most local sites fail here without knowing it.
- Put the answer in real text. Do not leave it only in an image, a PDF or a pop-up. If you cannot select the words with your mouse, a crawler may not see them.
- Give each service its own page and its own web address. One page called “Services” cannot match a question about one job.
- Remove walls. A login, a form gate or a slow page can stop a crawler before it reads a word.
- Link pages to each other. Every key page should be one or two clicks from your home page. Add an XML sitemap.
- Check that search engines have your pages. Use the URL Inspection tool in Google Search Console. DHM INTERPRETATIONOpenAI lists Microsoft among its search partners, so we also check Bing Webmaster Tools.
GOOGLE DOCUMENTEDA page must be indexed and eligible to show a snippet before it can appear in Google’s AI features. The same basics help with both. That makes this step worth doing, even if it seems routine.
Step 3 of 10Put the answer first
ChatGPT reads your page to answer a question. Make that easy. Put the answer in the first two sentences under each heading. Add the proof after it.
Aim for a block of about 100 to 150 words that makes sense on its own. If a stranger read only that block, would they get a full answer? If so, it can be quoted without the rest of the page.
For over a decade our family-run team has taken pride in serving homeowners across the region with dedication. Our approach to roofing begins with a conversation…
How long does a roof replacement take in Surrey? Most homes take two to three days. A big or steep roof takes longer. Rain can add a day. Here is how we plan it…
Example only. The wording is made up. Use your own real numbers.
Step 4 of 10Show who is behind the page
An AI tool cannot judge a business it cannot identify. Make the proof easy to find. Put it on the page. A profile alone is not enough.
- A named owner or author, with a short bio.
- Your licence, insurance or trade-body number, if your trade has one. Never invent one.
- Real photos of your own jobs, with a plain caption.
- Reviews from real customers, shown with their permission.
- Your address or service area, and your hours.
GOOGLE DOCUMENTEDGoogle values first-hand experience and a unique point of view over content that anyone could copy. OPENAI DOCUMENTEDOpenAI ranks results using several factors but has not listed them. Treat these steps as good practice, not a formula.
Step 5 of 10Build depth on each topic, not a pile of pages
One strong page per service beats twenty thin ones. For each service, build a small set of pages that support each other:
- One main page for the service.
- One page per area you really serve, each with details that are true for that area.
- A short question-and-answer section on the service page.
- Links between them, so a reader and a crawler can move around.
GOOGLE DOCUMENTEDGoogle advises against creating separate pages for every small change in wording and treats mass-produced pages as a spam risk. Our local landing page guide shows how to build area pages that are not copy and paste.
Step 6 of 10Say who you are the same way everywhere
Your business name, address and phone number should match on your site, your Google Business Profile and every directory. A mismatch gives any system a reason to doubt you. Our name, address and phone guide shows how to find the ones that differ.
Add an About page and a Contact page that say who you are, where you work and who you help. Plain words work best.
Structured data (schema) can describe all of this to machines. Be honest about what it does:
- GOOGLE DOCUMENTEDGenerative AI search needs no special markup. Keep using structured data for the rich results it already supports.
- OPENAI DOCUMENTEDThe requirement OpenAI lists for ChatGPT search is crawler access. The help page we read does not list schema.
- GOOGLE DOCUMENTEDMark up only what visitors can see on the page. Do not mark up fake or unrelated content. See Google’s general structured data guidelines.
So use schema to describe what is already on the page. Do not count on it to win a citation. Our guide on consistent business info for AI covers the checks.
Step 7 of 10Earn mentions on other sites
ChatGPT also reads the places that talk about you. INDEPENDENT RESEARCHYelp is the most used outside source for ChatGPT in its 2026 local study, and ties that to Yelp’s partnership with OpenAI. BrightLocal is a vendor, so read that as one data point, not a rule. We covered the full study in our post on why Maps rank and ChatGPT picks differ.
Real places to be mentioned:
- Yelp and your Google Business Profile, filled in and current.
- The directories your trade really uses.
- Supplier and manufacturer pages that list certified installers.
- Local groups, chambers, sponsorships and news stories.
Ask every happy customer for a review. Do not buy mentions. GOOGLE DOCUMENTEDFake or inauthentic mentions do not help, and we would not risk your name on them.
Step 8 of 10Write for the questions people really type
OPENAI DOCUMENTEDChatGPT uses an approximate location from your connection. A city, neighbourhood or postal code in the question helps it find local results. So name the places you serve on your pages, where it is true and natural.
Then find the real questions:
- List the five questions customers ask you on the phone.
- Check Google Search Console for long, chatty searches. Our earlier post has a free pattern for this.
- Answer each one on the page where it belongs, in the first two sentences.
Write one good answer per question. Do not make a new page for each way of asking it.
Step 9 of 10Add something only you can say
If your page says what ten other pages say, there is little reason to cite yours. Add what only a working tradesperson knows:
- Photos from your own jobs, with notes on what went wrong and how you fixed it.
- Local rules you deal with, like permits in your city. Check each one before you publish it.
- The mistakes you see other contractors make.
- Counts from your own work, like how many jobs you did last year, if you track them.
Keep every fact true and checkable. A made-up number does more harm than a blank space.
Step 10 of 10Measure it the honest way
Nothing in ChatGPT gives you a ranking report. So measure by testing. One try tells you almost nothing, because answers change from run to run. Use a sheet like this one.
| Question | Runs | Cited you | Named a rival | Note |
|---|---|---|---|---|
| Who can fix a roof in Surrey this week? | 5 | 2 of 5 | 4 of 5 | Cited on the leak page |
| How long does a roof replacement take? | 5 | 3 of 5 | 2 of 5 | Answer block worked |
| Is a roofer near me licensed? | 5 | 0 of 5 | 3 of 5 | Add licence proof |
- Pick five questions a real customer would ask, with your city in them.
- Run each one three to five times, on different days. Write it as “cited 3 of 5” and note how many runs you did.
- Check your analytics. Look for visits from chatgpt.com in your source report.
- Ask your host for crawler logs. Look for OAI-SearchBot visits.
- Remember: being cited is not the same as getting the call. Count your enquiries too.
DHM DATASource: our Google Analytics 4 property and our live robots.txt. A small site, so directional only.
We share our own numbers on purpose. For a small site, ChatGPT is a small source of visits today. Set your hopes by what you can measure, not by what a vendor promises.
Go deeper
More on how ChatGPT finds local businesses
Each one covers a piece of the guide above.
How ChatGPT finds local businesses
The sources it reads and how a local name gets in.
See the guide › Guide 02Is your business invisible to ChatGPT?
The first check to run, and the quick fixes after it.
Read the post › Guide 03How AI Mode picks its citations
The same job on Google’s side of the fence.
Read the post ›What are the most common mistakes?
- Blocking by accident. An old Disallow line, or a bot shield at your host, keeps the crawler out.
- Hiding the answer. The facts sit in an image, a PDF or a pop-up.
- One page for everything. A single Services page cannot match a specific question.
- Copy-paste city pages. They add pages and no proof.
- Fake reviews or fake mentions. They put your whole name at risk.
- Relying on schema alone. It describes the page. It does not replace a good page.
- Checking once. One run proves nothing. Answers change.
- Mixing up a citation and a customer. Track calls and booked jobs too.
How long does it take?
OPENAI DOCUMENTEDA robots.txt change can take about 24 hours to take effect for search. That covers access only. OpenAI gives no date for when new or improved pages get cited, and neither can we.
DHM INTERPRETATIONWe recheck our test questions at weeks one, two, four and eight. If a page is never cited after that, we fix the page, not the clock.
Your 14-point checklist to do this week
- Crawler allowedOAI-SearchBot is allowed in robots.txt.
- Host and CDNOpenAI’s IP ranges can reach the site.
- No stray blockNo old
Disallow: /line is left over. - Real textKey pages are HTML text, not images or PDFs alone.
- IndexedKey pages show in Google and Bing.
- One page per jobEach service and main area has its own page.
- Answer firstEach page answers its main question in two sentences.
- Named peopleAn owner or author is named, with proof.
- Same factsName, address and phone match everywhere.
- About and ContactThey say who you are and where you work.
- Honest markupStructured data matches visible text only.
- Profiles currentYelp, Google Business Profile and trade directories are right.
- Original proofKey pages hold a photo, job note or local rule only you have.
- A test sheetFive questions, run three to five times each, saved with dates.
See your Maps side first.Before you pay us a dollar, we run a free 49-point geo-grid scan of your service area. You keep the results either way. The AI side you can check by hand with the sheet above.
Get the free scanWhat this guide can’t tell you
- The full ranking recipe is not public. OpenAI lists the crawler rules and the eligibility step. It does not list the ranking factors.
- We ran no controlled test. Our steps rest on OpenAI’s and Google’s published pages, on one vendor study, and on our own small data.
- Pages change. We read the OpenAI and Google pages on November 19, 2026. Check them again before you act.
- The pictures are illustrations. The ChatGPT screen is a mockup, not a real reply.
- “Cited” is not “chosen.” A link in an answer is not a booked job.
For a worked example in one trade, what an assistant checks before it names a tow company breaks one spoken question into the five things it looks up.
We do not guarantee that ChatGPT will cite your business. Nobody honest can. We take one client per trade, per city. What we can do is make sure your site, your profiles and your proof give it something real to find.
How do I get ChatGPT to cite my website?
Let OpenAI’s search crawler, OAI-SearchBot, read your site. Then give it a page worth citing: a clear answer, proof of who you are and matching details across your other profiles. OpenAI says placement in ChatGPT answers is not guaranteed, so test and keep improving.
Do I have to allow OAI-SearchBot to show up in ChatGPT search?
OpenAI says sites that opt out will not be shown in ChatGPT search answers, though they can still appear as plain links. OpenAI recommends allowing OAI-SearchBot in robots.txt and letting its published IP ranges through your host or CDN.
Does blocking GPTBot keep me out of ChatGPT search?
No. OpenAI says its crawler settings are independent. GPTBot is tied to training OpenAI’s models; OAI-SearchBot is tied to ChatGPT search. You can allow one and block the other.
How long does a robots.txt change take to work?
OpenAI says its search systems may take about 24 hours to adjust after a robots.txt update. That covers crawler access. It does not set a date for when, or whether, a page gets cited.
Does schema markup help ChatGPT cite me?
OpenAI’s help page names crawler access as a requirement and does not list schema. Google says generative AI search needs no special markup. Use structured data to describe what is visible on the page, but do not count on it to win a citation.
Can anyone guarantee that ChatGPT will cite my business?
No. OpenAI says ChatGPT ranks results using several factors and that placement is not guaranteed. Anyone who promises a citation or a date is guessing. We do the work that makes a citation possible, then we test.
Sources
- OpenAI, “Overview of OpenAI crawlers.” Roles of OAI-SearchBot, GPTBot and ChatGPT-User, the independent settings, IP ranges and the 24-hour note. Retrieved November 19, 2026.
- OpenAI Help Center, “Searching the web with ChatGPT.” Query rewriting, location use, ranking and the note that placement is not guaranteed. Retrieved November 19, 2026.
- Google Search Central, “Optimizing for generative AI search.” Page last updated July 10, 2026. Retrieved November 19, 2026. This is Google’s guide for Google’s AI features, not for ChatGPT.
- Google Search Central, “General structured data guidelines.” Page last updated July 10, 2026. Retrieved November 19, 2026. Written for Google rich results.
- BrightLocal, “AI local search study” (Mike Hawkes, published September 16, 2026). Vendor research. BrightLocal sells the AI tracker it used.
- BrightLocal, “Where to get local citations for AI search” (published August 26, 2026). Source for the Yelp and ChatGPT point.
- Digital Handshake Media, our live robots.txt, checked November 19, 2026, and our own Google Analytics 4 property, August 21 to November 19, 2026. A small site, so directional only.
Next step
Find out whether ChatGPT can read your site
One client per trade, per city. The call starts with that check, not a pitch. Free, and you keep the findings either way.
