ChatGPT knows about a new product only after one of its retrieval paths has read the product page. A product that went live on your site this week is not automatically in ChatGPT's answers. OpenAI runs separate crawlers for training and for search, and a new product reaches each one on a different schedule. If the search crawler has not visited the page, if the page is not linked from anywhere the crawler follows, or if ChatGPT is answering from remembered training data instead of live search, your new product stays invisible while older products from competitors keep showing up.
There is no form to submit a new product to ChatGPT. Nobody can place it there for you. What you control is whether the search crawler can reach the page, whether the page is easy to discover, and whether the facts on it are plain enough to be quoted.
Why ChatGPT misses new products
ChatGPT fails on new products in ways that have little to do with product quality:
- Training and search are separate channels. GPTBot gathers content that may be used to train models. OAI-SearchBot builds the index behind ChatGPT search answers. Allowing one does not allow the other, and blocking GPTBot does not keep a product out of search answers. A new product will not appear in answers built from training data until a newer model arrives, but it can appear in search answers as soon as OAI-SearchBot has crawled the page.
- OAI-SearchBot has not visited yet. A new page with no links pointing to it, a page missing from the sitemap, or a page behind a block the team forgot about will not be found on the crawler's own schedule. OpenAI says it can take about 24 hours from a robots.txt update for its search systems to adjust, so even a fix needs a day to take effect.
- The retrieval mix is blended and undisclosed. OpenAI's help pages say ChatGPT search sometimes partners with other search providers, and the privacy policies linked there name Microsoft. A product page that exists only in Google's index and nowhere else can be missed by some of ChatGPT's retrieval paths.
- The answer-time fetch does not index anything. ChatGPT-User visits a page when a person asks ChatGPT to read it. That fetch helps one conversation. It does not put the product into the search index for the next person who asks.
Step 1: Allow OAI-SearchBot, and check the edge
Check your robots.txt at [yourdomain.com]/robots.txt, replacing yourdomain.com with your actual domain. If nothing loads, create the file at the root of your site. Then add a group for the search crawler near the top, before any broad disallow rules:
User-agent: OAI-SearchBot
Allow: /
Do not stop at the file. The file states what you intend, and the edge decides what really happens. A firewall or bot-protection setting can block OAI-SearchBot even when robots.txt allows it. In Cloudflare, open Security, then Bots, and confirm the OpenAI crawlers are not being challenged or blocked. Then verify in your own logs, and confirm the requests are genuine against OpenAI's published IP ranges for the search bot:
grep -i "oai-searchbot" /var/log/nginx/access.log
No hits over an extended period can mean the crawler has not returned yet, or something upstream is still blocking. Check both possibilities, regardless of what the file says. Leave GPTBot out of this step unless you have a separate reason to change its setting. The two settings are independent, and changing the training crawler does nothing for a new product this week.
Step 2: Make the product page discoverable
A crawler finds pages by following links and reading sitemaps. A new product page that nothing links to is easy to miss.
- Link the new product from its category page and from any related product pages, using the product name as the link text.
- Add the product URL to your XML sitemap, and confirm the sitemap itself is listed in robots.txt or submitted where your search tools accept it.
- Keep the product URL stable. A URL that changes with every promotion fragments the index presence of the page.
- Do not hide the product behind a login, a geo gate, or a script that only loads after a shopper clicks. If a crawler cannot reach the page on its own, the product does not exist for search answers.
Step 3: Put the product facts in plain text
Write the product name, what it is, what it costs, and whether it is in stock as visible text on the page, not only inside images or scripts. ChatGPT quotes what it can read. A page that shows the facts only after JavaScript runs is a weaker source than a page that states them in the HTML, because OpenAI publishes no claim that its crawlers render JavaScript. The structured data on the page should state the same facts as the visible text, so a reader that takes the markup and a reader that takes the page agree with each other.
Step 4: Feed the Bing side of the retrieval mix
OpenAI's help pages name Microsoft among its search partners, and Bing remains one of the retrieval sources behind ChatGPT's live answers. Register the site in Bing Webmaster Tools and confirm the new product pages are indexed there. Then set up IndexNow, which lets your site tell participating search engines the moment a URL is added, updated, or deleted, so Bing hears about the new product instead of waiting for the next crawl. IndexNow is a notification, not a guarantee of indexing, and it does not cover Google, so keep your Google steps separate. This step is specific to ChatGPT and Copilot.
Step 5: Test the way a customer would
Ask ChatGPT the question a customer would ask, with web search on, and read the citations. If it names a competitor, check whether the competitor's product page is indexed and yours is not. If it names your old product instead of the new one, the old page is the one the index holds, and the new page still needs the steps above. If it cites a reseller or a marketplace listing for your new product, that listing is currently the strongest source ChatGPT can find, and your own product page needs to become the stronger one. Retest after a recrawl, not after every edit, and log what was cited each time so you can see whether the fix landed.
A hypothetical, to make it concrete
Say you run a trail running store and just launched a new waterproof shoe. The product page is live, but it is linked only from the homepage banner, it is missing from the sitemap, and an old robots.txt group that was meant to block training also blocks OAI-SearchBot. A scan would flag it in one pass: search crawler disallowed, new page undiscoverable, no IndexNow notification sent. The fixes run in this order: allow OAI-SearchBot and confirm the edge lets it through, link the shoe from its category page and add it to the sitemap, put the price and stock status in plain text on the page, and set up IndexNow so Bing hears about the new URL. The training-based answer still waits for a newer model, but the search-based answer can change after the next crawl.
The fundamentals
These mechanics are shared across every engine and covered in full in their own guides:
- •llms.txt: the file format, a complete sample, and per-platform publish steps, see our guide to writing a good llms.txt file
- •Product schema: the JSON-LD markup that states price and availability in structured form, see our product schema guide for AI search
- •Every AI crawler in one reference table, see our robots.txt guide for AI search
What this guide does not cover
Getting a new product known is the start. Prices change, products sell out, and crawler behavior changes, which is why monitoring exists. And being known is not being bought from: letting the agent complete the purchase is a separate job.
Check your own brand now
Run the free scan. It takes seconds, no signup, and shows exactly where ChatGPT stands with your site: aizii.ai/aeo-scan
Run the free scanSources
- OpenAI bots documentation (specific): https://platform.openai.com/docs/bots
- OpenAI help: searching the web with ChatGPT (specific): https://help.openai.com/en/articles/9237897-searching-the-web-with-chatgpt
- IndexNow (specific): https://www.indexnow.org
Note: links marked specific point to the page that supports the claim in this guide. Links marked general docs point to the vendor's general docs section for that topic, not to a page that proves every step. Check the vendor docs for the current path before relying on a menu name or click path in this guide.
