Crawl Nine

What are the ChatGPT Ads landing page requirements?

Published · Last updated

Sources last checked . This page describes a product that changes often. Every claim is linked to the page it came from, so you can check any of it yourself.

What are the ChatGPT Ads landing page requirements?

Four things. The page must be a working web address that loads for anyone. It must not block OpenAI’s automated readers. Its content must comply with OpenAI’s ad policies, not just the ad’s. And it must be consistent with the ad pointing at it. OpenAI may visit the page to check it before the ad runs.

There is a fifth thing that is not a rule but decides your results: OpenAI reads the page to help work out when your ad is relevant. So the page is part of your targeting, not just your destination.

Every quotation below is from OpenAI’s own advertiser documentation, read on 4 October 2026. Where we add our own judgement, the page says so.

Does the page have to let OpenAI’s crawler in?

Yes, and this is the requirement that costs the most money because it fails silently. OpenAI instructs advertisers to “Ensure landing pages are valid and reachable: all landing pages must use valid links and must not block OpenAI user agents (OAI-AdsBot and OAI-SearchBot)”.

Two agents are named there, but OpenAI’s crawler guidance makes only one compulsory for the landing page. OpenAI: “You must allow OAI-AdsBot. We recommend allowing both OAI-AdsBot and OAI-SearchBot.” If you advertise from a product feed, a list of your products uploaded to OpenAI’s Ads Manager, there is one more rule: “For product feeds, also allow OAI-SearchBot to crawl the product image URLs.” That means wherever your product photos are stored. OpenAI warns that if that reader is turned away there, it can stop your feed being processed.

OpenAI gives the wording in a sentence. Set out the way it has to appear in the file, it is four lines:

User-agent: OAI-SearchBot
Allow: /

User-agent: OAI-AdsBot
Allow: /

Robots.txt is only the first of three layers. OpenAI also names “Web protection and bot mitigation”, Cloudflare, Akamai and similar, and application-level checks such as CAPTCHAs and JavaScript challenges. Any of the three can stop the reader on its own.

Asking support to wave it through is not a plan. OpenAI’s own answer to that question is: “Do not rely on a manual bypass.”

The mechanics of that reader have their own page here: what is OAI-AdsBot.

What kind of page can you send people to?

An ordinary web page you control. OpenAI warns that “App-store links, deep links, documents, or destinations that require an app, login, region-specific access, or unsupported redirects may not provide enough crawlable content for validation or review”, and advises using “a directly reachable web landing page whenever possible”.

That rules out several things businesses reach for by habit. A link straight into the App Store. A PDF brochure. A page behind a login. A page that only opens for visitors in one country.

Redirects deserve their own mention. A chain of them gives the reader more chances to be stopped and gives you more places for a mistake to hide. Point the ad at the final address.

The practical test is simple and you can run it without any tools: open the address in a private browsing window, on a phone, signed out. If you see the page, so can a person who clicks your ad. Whether the automated reader sees it is a separate question, and that one needs an actual check.

Does the page content have to follow the ad rules?

Yes, and end to end. OpenAI states that “your ad creative, image, and landing page must comply with OpenAI’s Ads Policies”, and that “These standards apply to all ad assets, including copy, images or video, and landing pages.”

The consistency rule is the one that catches people out, and OpenAI gives its own example: “Ads must be consistent end-to-end: approved creatives may not link to destinations that introduce disallowed content (for example, a food delivery ad linking to alcohol delivery).”

So an approved ad pointing at a page that sells something excluded is not an approved campaign. It is a rejected one waiting to be found.

If you sell health or financial services, the page is where the licence claim lives too, OpenAI notes that “Advertisers may be required to provide proof of licensure.” Legal services carry their own rule: US ads are permitted only where the advertiser is licensed in the jurisdiction the ad is shown. Which categories are open to whom is on which businesses can advertise on ChatGPT.

There is also a rule about not looking like ChatGPT: “Ads must be clearly distinguishable from the ChatGPT product experience.”

Why does page quality matter more here than on Google?

Because there is no keyword to bid harder on. OpenAI states that ad selection “considers multiple signals, including the context and intent of the current conversation, the ad’s landing page, title, copy, advertiser-provided context hints…”, the list runs on past that. The page is an input to whether your ad shows at all.

On Google you can compensate for a weak page by bidding harder on a keyword. Here there is no keyword to bid on. Context hints are the closest thing. OpenAI says they help guide the match but are not exact-match keywords and do not guarantee delivery. You can upload a customer list, OpenAI calls these custom audiences, but that decides who is allowed to see the ad, not which conversations it turns up in, and it needs at least 25,000 matched people before you can target them at all. Details on how ChatGPT Ads targeting works.

The auction is “a relevance-weighted, second-price auction”, so relevance is not a tiebreak, it sits alongside your bid in deciding who shows.

This is the part that changes what an agency should actually do for you. Most of the work is not in the campaign, it is in the page: making it fast, making it readable without JavaScript, making it obviously about the thing the ad promised.

Does the page need to work without JavaScript?

It is not written as a requirement, and it is still the safer assumption. Research from Vercel and MERJ published 17 December 2024 found that the OpenAI, Anthropic, Meta, ByteDance and Perplexity crawlers did not run JavaScript. It named two that do: Google’s Gemini and Apple’s crawler. A page whose text only appears after scripts run may be a blank page to most of that list.

This is the one claim on this page that is not from OpenAI, and it is flagged rather than blurred because that matters. The study is from December 2024 and OAI-AdsBot did not exist then. OpenAI has published nothing about whether its ads reader runs JavaScript, so this is an inference from how its sibling crawlers behaved, not a fact about this one.

What we can say is what the requirement rests on. OpenAI says it “may also use content from the landing page” to judge relevance. If that content does not exist until a script has run, there is a real chance there is nothing to use.

The cost of being wrong is asymmetric. Building a page whose words are in the HTML is ordinary work with no downside. Building one that needs JavaScript and hoping the reader runs it risks the whole campaign on an assumption nobody has published.

What breaks most often, in practice?

A security service turning the reader away. OpenAI names it directly: services such as Cloudflare and Akamai “can mistakenly block legitimate crawlers, often returning 403 Forbidden errors”. The site works perfectly for people, so nothing on your screen says anything is wrong.

The symptom appears in the ad tool rather than on your website. OpenAI: if an ad’s status reads “Not serving”, the common causes include “landing pages that block OpenAI user agents or IP addresses”.

Fixing the block may not be enough on its own, OpenAI notes ads “may need to be re-uploaded or resubmitted for review after the landing page becomes accessible.”

A short checklist worth handing whoever runs your site, taken from OpenAI’s own guidance: does the page return a successful response to OAI-AdsBot; does robots.txt allow that path; and do the firewall, CDN, bot mitigation, JavaScript challenges, CAPTCHAs, authentication or geo rules block automated access.

Check it before you spend anything. Our free crawler check tries the agents for real against your address and shows what came back, with no email and no account. If the page needs more than unblocking and you want us to run the ads, Amplify, our ChatGPT Ads management, checks the landing page and makes minor changes within the agreed scope. A substantial new page is quoted separately.

Sources

All fetched 4 October 2026, except where dated otherwise.