Why doesn’t ChatGPT suggest my plugin?
You fixed your listing, your plugin shows up in directory search, and ChatGPT still never brings it up when someone asks for help. We wanted to see how ChatGPT picks the plugins it suggests in a conversation, so we asked it to search for plugins and tell us exactly what its search returned.
The listing work that lifts a plugin in directory search turned out to make no difference to whether ChatGPT can suggest it. What does, OpenAI says, is how many people install a plugin and keep using it.
What a plugin suggestion looks like
When ChatGPT decides a plugin could help, it searches for one with a query it writes itself. It often adds brand names it already knows: for a question about CRM deals, it searched for “CRM Salesforce HubSpot deals opportunities”. Then it shows a card titled “ChatGPT needs to use a plugin” with one to five plugins in order, each with Install and Not now buttons.
The card comes up most when people ask which plugin to use. Asked to do a task, ChatGPT more often does the job itself with web search, code or image generation, or names a plugin in the text of its answer.
Which ChatGPT plan and mode show plugin cards
The mode matters even more than the question. We ran the same requests on a free account and on a paid one.
- Free plan, default mode. ChatGPT showed a plugin card in about 2 of 50 answers. The rest of the time it named apps in its reply, searched the web or did the job itself. The plugin tool is there: when we asked it outright to search and suggest a plugin, a card appeared.
- Paid plan, thinking mode. About 15 of 50 answers came with a card, often with several plugins in order.
Most people use ChatGPT on the free plan in the default mode. For them a plugin shows up, if at all, as a name in the reply, without an Install button next to it.
On any plan, a card needs a search result first. So we looked at the search.
The test: asking ChatGPT to find plugins by name
We picked plugins at random from the directory, asked ChatGPT to search for each one by its exact name, and then asked it to list every result its search returned. If ChatGPT can suggest a plugin, a search for its exact name should find it.
For most of them the search came back empty, or returned other plugins with a similar word in their names. The plugin we asked for wasn’t there.
Top-ranked plugins were missing too
The misses weren’t only small or new plugins. Plugins in the top 5 of their category, shown on the directory home page, didn’t come back either.
And when ChatGPT does suggest plugins, it doesn’t pick the directory’s leaders:
| Question to ChatGPT | ChatGPT suggested | Their directory position |
|---|---|---|
| Which plugin can compare hotels? | Expedia, Tripadvisor, Booking.com | #23 and #27 for “hotels”; Booking.com #1 for “booking” |
| Show my CRM deals likely to close | Attio, HighLevel, Close | #19 to #21 for “crm” |
The directory’s top plugins for “hotels” and “crm” stayed out of the card. The card doesn’t follow the order you see in directory search.
The same names, whatever the topic
Broad searches gave the clearest picture. They return dozens of plugins, and the same names keep coming back. A search for EV charging stations returned data tools and e-signature apps, and no charging plugin.
Within that set, the search matches by meaning. A search for “AI notetaker for in-person meetings” returned Otter.ai, Read AI, Fireflies, Granola and Fathom. Outside it, there is nothing to find.
ChatGPT’s plugin search covers a limited pool. By our estimate it holds a few hundred of the more than 5,000 plugins in the directory, and a plugin outside the pool never appears in a suggestion card.
How ChatGPT decides which plugins to suggest
A developer asked OpenAI on X how plugins get recommended in conversations. Max Stoiber, who works on plugins at OpenAI, answered:
Oct 9, 2026 · View on X“ranking across the directory and in-chat suggestions is fully deterministic …”
- conversion
- active usage
- retention
Conversion is how many of the people who see your plugin install it. Active usage is how often they call it, and retention is whether they keep coming back. The same three factors rank plugins in directory search and in chat. His advice: make something people want to use again.
That fits our tests. The plugins that came back for almost any search belong to big brands with millions of customers, like Tripadvisor, Salesforce and Calendly. Most plugins we tried didn’t come back even for their own name.
So nobody picks the pool by hand, and a better name or tagline alone won’t put you in it. OpenAI’s rules for plugin developers point the same way. Plugins with strong real-world utility and high user satisfaction:
“may be eligible for enhanced distribution opportunities, such as directory placement or proactive suggestions”
OpenAI, plugin guidelines
Check if ChatGPT can suggest your plugin
- Start a new chat in thinking mode, which follows the instructions most closely.
- Send:
Search the ChatGPT plugin directory with exactly this query: "Your Plugin Name". Use only that query, and don't suggest or install anything. - Then send:
What exactly did your plugin search return? List every plugin from the search results in the exact order you received them. One per line, numbered, names only.
If your plugin comes first, it’s in the pool. An empty list, or a list of other plugins, means ChatGPT can’t suggest yours yet. Run the same check for two or three competitors to see who in your area is in.
How to get your plugin into the pool
Work on the three things OpenAI ranks by. All three come from real users, so no listing trick moves them.
- Conversion: turn views into installs. Your listing page decides whether people install. Show the job in the name, tagline and example prompts, and say what the plugin connects to and what it can change. Link to it from your app, onboarding emails and help docs: your own customers are the easiest installs.
- Active usage: make the first run work. A failed sign-in or an error on the first call loses the user before they see any value.
- Retention: own one frequent job. People come back every week to a plugin that does one job they need often. One that does many things once gets tried and forgotten.
Repeat the check every few weeks to see when your plugin gets in.
Once you’re in: how to get picked
- Describe the job in plain words. ChatGPT matches plugins by meaning, so say what the plugin does in the words your users would type.
- Make your brand known outside ChatGPT. ChatGPT adds brands it knows to its search query, and for “which plugin” questions it often checks the web first, including OpenAI’s own plugin pages.
- Judge by repeated checks. ChatGPT writes a new search query each time, so the same question can bring a different card. Look at how often you appear across several runs.
What to do in the meantime
Most people who find a plugin today find it in directory search, and directory search is open to every plugin. Put the words your users search into your name and tagline, then track your positions. Start with why your ChatGPT plugin isn’t showing up in search.
FAQ
Does ChatGPT recommend plugins in conversations?
Yes. When ChatGPT decides a plugin would help, it shows a card titled “ChatGPT needs to use a plugin” with one to five plugins and an Install button for each. It picks them from a limited pool, not from the whole directory.
How does ChatGPT rank plugins?
By three factors, according to OpenAI: conversion, active usage and retention. The ranking is deterministic and applies both to the plugin directory and to suggestions in chat.
Does ChatGPT suggest plugins on the free plan?
Rarely. On the free plan in the default mode, ChatGPT showed a plugin card in about 2 of 50 answers in our tests and named apps in its reply instead. In thinking mode on a paid plan, about 15 of 50 answers came with a card.
Why doesn’t ChatGPT suggest my plugin?
Most likely your plugin isn’t in the pool ChatGPT searches when it suggests plugins. By our estimate the pool holds a few hundred of the more than 5,000 plugins in the directory. OpenAI ranks plugins by conversion, active usage and retention, and plugins with few users don’t come back even when ChatGPT searches for their exact name.
How do I get my ChatGPT plugin suggested in conversations?
Get more people to install it and keep using it. OpenAI ranks in-chat suggestions by conversion, active usage and retention. Make the listing convert, make the first run work, and check every few weeks whether ChatGPT’s plugin search finds your plugin by name.
Does a higher rank in the ChatGPT plugin directory get my plugin suggested?
Not by itself. ChatGPT’s suggestion cards don’t follow directory search positions, and plugins at the top of their category can be missing from its plugin search. Both rankings reward the same things, OpenAI says: installs, active use and retention.
How can I check if ChatGPT can suggest my plugin?
In a new chat, ask ChatGPT to search the plugin directory for your plugin’s exact name, then ask it to list every result its search returned. If your plugin comes first, it is in the pool.