Custom GPTs retire on December 11: what happens to your GA4 reporting GPT

4 min read
A GA4 reporting GPT being migrated to a plugin: instructions and knowledge files move, the GA4 custom action does not. A note reads Custom GPTs retire Dec 11, rebuild the data step before then.

OpenAI is retiring custom GPTs. The standard date is December 11, 2026. If you built a GPT that pulls GA4 numbers for your client reports, the part that fetches the data is the part that does not survive the move.

What OpenAI announced

Custom GPTs are scheduled to retire on December 11, 2026. FedRAMP workspaces and Enterprise workspaces with an approved deferral get until February 11, 2027; that later date does not apply to every workspace. On the retirement date, custom GPTs and their GPT pages become inaccessible. OpenAI says to follow the notice for your own account or workspace, which can carry a different timeline.

The replacement is a plugin. When migration is available for your account, My GPTs shows a "Migrate to plugin" option for each GPT you created.

What moves over, and what does not

According to OpenAI's migration FAQ:

  • Moves: your instructions (they become a skill in the plugin), your knowledge files (copied into the plugin's reference files) and any connected apps.
  • Does not move: custom actions, the GPT's selected model and your sharing settings. Your existing conversations stay with the original GPT; OpenAI says they remain accessible after retirement.

Migration also uses the GPT's latest published version. Unpublished edits are left behind, and the original GPT becomes read-only once you migrate it.

Why a GA4 reporting GPT is the one that breaks

A GPT that only rewrites numbers you paste in will migrate cleanly. A GPT that fetches live GA4 data through a custom action will not, because custom actions do not transfer. In OpenAI's words, features that depend on them "will not work in the plugin until you set up a replacement," and rebuilding the connection "may require a custom MCP server and technical setup."

So the instructions and the report format come across. The step that gets the data does not.

If your clients use the GPT too

Migration does not give a GPT's existing users access to the replacement. A migrated plugin starts private, it is not installed for anyone automatically, and making it public is a separate submission process. If you shared a reporting GPT with clients or teammates, each of them needs access to the new plugin and has to install it before December 11.

What to do before December 11

  1. List the GPTs your reporting depends on, and note which ones you created and which ones someone else maintains.
  2. For each one you created, check whether it uses a custom action. That is the part you will have to rebuild.
  3. Publish any edits you want to keep before you migrate. Drafts do not transfer.
  4. Save a few prompts you run every month, so you can compare the plugin's output with the GPT's.
  5. Decide who has to switch, and confirm each of them can install the replacement.

Three ways to keep the report going

Rebuild the action yourself. Migrate the GPT, then replace the custom action with an MCP server that calls the GA4 Data API. It is the most control, and it is now software you host, secure and keep running.

Use a plugin that already connects to the data. Plugins can bundle a connected app, so the data connection becomes someone else's job to maintain.

Take the report out of the chat. If the GPT's job was to produce the same monthly summary for each client, it may not need a chat at all. MetricsWatch connects to GA4 once per client and emails the report on a schedule, with your branding, so nobody has to open ChatGPT and ask for it.

Whichever you pick, set it up now. OpenAI says to migrate and test before the retirement date, and to allow extra time for GPTs that use custom actions.

custom gpt retirement chatgpt plugins ga4 reporting client reporting

If the GPT's job was the same monthly GA4 summary for every client, a scheduled report does it with no GPT to maintain.

Ready to streamline your reporting?

MetricsWatch pulls twelve sources into one white-label report, delivered on your schedule: Google Analytics, Google Ads, Google Search Console, Google Business Profile, Google Sheets, Microsoft Advertising, Facebook Ads, Facebook Pages, Instagram, LinkedIn Ads, LinkedIn Pages, and Mailchimp. Starts at $49/month. 14-day free trial, no credit card required.

Related Articles

MetricsWatch vs TapClicks comparison card

MetricsWatch vs TapClicks: Report Sender or Marketing Data Platform?

TapClicks is a marketing data platform with 250+ connectors, dashboards, AI agents and scheduled reports; MetricsWatch emails the finished client r...

MetricsWatch vs Geckoboard comparison card

MetricsWatch vs Geckoboard: Client Reports or Team Dashboards?

Geckoboard builds live dashboards for teams, on office TVs, Slack and shared links; MetricsWatch emails the finished client report in the body of t...

MetricsWatch vs DashClicks comparison card

MetricsWatch vs DashClicks: Report Sender or Agency Platform?

DashClicks is an all-in-one agency platform with a client dashboard and resold fulfillment; MetricsWatch emails the finished marketing report in th...