In this guide
- An assistant connected through the Tag Manager API or an MCP server can read containers and create tags, triggers and variables in a workspace.
- Grant the least permission that does the job, review every change and publish yourself.
- Keep a backup export and use versions with notes, so any change can be undone.
The most useful thing an AI assistant can do for a marketer is the boring part of Tag Manager: build twelve tags with the right names, triggers and consent settings. The risk is just as clear: an assistant that publishes a wrong change to a live container can break tracking on a live store. This page explains what is safe to delegate, which permissions to give and a workflow that keeps you in control. For a hands-on set-up see the Open Tracked MCP server guide and MCP for Google Tag Manager.
How the connection works
Assistants such as Claude, ChatGPT or Cursor talk to Tag Manager in one of two ways. Through the Tag Manager API, which needs OAuth consent for specific scopes. Or through an MCP server, a small program that exposes Tag Manager actions as tools the assistant can call, using the same API underneath. In both cases the assistant acts with your Google account's permissions on the containers you allow.
Ask Ari instead of digging through reports
Ari reads your tracking and ad data and answers in plain words.
Permissions: give the least
| Scope or role | What it allows | Give it for |
|---|---|---|
| Read-only | View containers, tags, triggers, variables | Audits, explanations, finding problems |
| Edit containers (workspace) | Create and change items in a workspace | Building and fixing tags |
| Edit container versions | Create versions from a workspace | Saving a reviewed state |
| Publish | Make a version live | Almost never. Do this yourself |
If you can, create a separate Google user for the assistant, with access only to the containers it needs. Avoid giving it admin on the whole account. You can always raise permissions for a task and lower them afterwards.
What it does well
Audits
Read a container and list duplicate tags, tags with no consent setting, unused triggers and naming problems. See Google Tag Manager audit.
Bulk building
Create a set of event tags for GA4 or Meta with the right names and triggers.
Explaining
Describe what a container does in plain words for a handover.
04Drafting server-side set-ups
Prepare the server container pieces for a platform, ready for you to review.
Connected in minutes
Add your domain, install the pixel, connect your ad platforms and watch the first events land.
What it should not do alone
- Publish. A wrong publish goes live to every visitor.
- Delete containers or users. Hard to reverse.
- Change consent logic without review.
- Send data to new vendors. That is a privacy decision.
- Handle secrets such as access tokens, beyond what it is given for the task.
A safe workflow
Export a backup
Download the container as JSON before the session.
Create a new workspace
Have the assistant work only in that workspace.
Give a clear task
Say what to build, with names, triggers and consent settings.
Review in Tag Manager
Open the workspace changes list and read each item.
Test in Preview
Check that tags fire as expected.
Create a version with notes, then publish yourself
After you are satisfied.
Record what happened
Ask the assistant to list every change it made and keep that list with the version notes. If something goes wrong a month later, the record tells you where to look. Keep the conversation or a summary of it with the container notes.
Privacy
An assistant reads what is in your containers, which may include IDs and sometimes tokens in variables. Remove secrets from variables where you can, and use an assistant and a connection you trust. Do not paste access tokens into chats. See data loss and leakage prevention.
Orders from your store, back to the ad that earned them
Your store sends the order once. Open Tracked passes it on to each ad platform, cleaned and deduplicated.
Frequently asked questions
Can an AI assistant control Google Tag Manager?
Yes, through the Tag Manager API or an MCP server, with permissions you grant.
Should AI publish my container?
No. Keep publishing for yourself after review and testing.
What permission should I give?
The least that does the task: read-only for audits, workspace edit for building.
How do I undo a change?
Restore a previous version or import your backup export.
Is it safe to give an assistant my tokens?
No. Keep secrets out of chats and variables where you can.
Do I need spreadsheets to read my results?
No. Once tracking is live, the Open Tracked analytics shows spend, revenue, ROAS and profit, and you can ask Ari, the built-in AI assistant, questions about your own numbers in plain words. Ari's allowance depends on your analytics plan.
Is Open Tracked different from Open Track or Open Space Track?
No. Open Tracked is the official name of our server-side tracking platform, and the website is opentracked.io. Many people search for it as Open Track, OpenTrack or Open Space Track. All of them lead to the same product: server-side Google Tag Manager hosting, first-party analytics and a CRM pipeline.
Fix your tracking at the source
Open Tracked hosts your server-side Tag Manager, keeps first-party cookies alive and shows what each ad is really worth.
The Open Tracked team
We build Open Tracked: server-side tracking hosting, analytics and the tools around them, for stores that spend on ads. Questions about this guide, or an idea for the next one?
Comments
Loading comments...
Integrations · 5 min read
