In this guide
- A good prompt states the platform, the goal, the constraints and the output format.
- Give the assistant facts, not secrets. Paste event names and error messages, not tokens or personal data.
- Check what it says against the platform's documentation and Preview. Assistants can be confidently wrong.
AI assistants are good at tracking work that is mostly reading, structure and checklists: they draft a data layer spec, explain an error, compare two containers. They are poor at knowing your exact set-up unless you tell them. This page gives prompts you can copy and adapt, and rules for using them safely. It goes with AI controls for Tag Manager and the Open Tracked MCP server guide.
Rules for better answers
- Say what platform and version. "Server-side Google Tag Manager" and "Meta Conversions API" are not the same as "GTM".
- Give the goal. "I want purchases counted once in Meta" beats "fix my tracking".
- Paste the evidence. Error text, the event data from Preview, the list of tags.
- State constraints. "No new vendors", "must respect consent".
- Ask for a format. A table, a checklist, a step list.
- Ask for assumptions. "List what you are assuming."
Connected in minutes
Add your domain, install the pixel, connect your ad platforms and watch the first events land.
Fifteen prompts
| Job | Prompt (adapt the brackets) |
|---|---|
| Data layer spec | "Write a data layer specification for a [Shopify] store covering view_item, add_to_cart, begin_checkout and purchase, with field names, types and an example push for each. Use GA4 names." |
| Container audit | "Here is the list of tags, triggers and variables from my container: [paste]. Find duplicates, missing consent settings, unused items and naming inconsistencies. Return a table with severity." |
| Debugging a failed tag | "My server container Meta tag shows failed. The outgoing request and response are: [paste]. Explain the likely cause and what to check next." |
| Event mapping | "Map these GA4 events to Meta, TikTok and Google Ads standard events in a table: [list]." |
| Consent check | "List which of these tags should require advertising consent and which analytics consent, with reasoning: [list]. Do not give legal advice; flag what I should confirm with a lawyer." |
| Deduplication plan | "Explain how to share one event ID between a browser pixel and server tag in GTM for purchases. Give steps and what to check in Meta Test Events." |
| Request volume estimate | "I have [N] monthly visits and send these events per visit: [list]. Estimate monthly requests and which hosting plan fits, with assumptions." |
| GA4 naming review | "Review these GA4 event names and parameters against GA4 naming rules and recommended events: [list]. Suggest renames." |
| Hashing check | "Show the steps to normalise and SHA-256 hash e-mail and phone for the Meta Conversions API, with an example in [language]." |
| Test plan | "Write a test plan for a new server-side Meta Conversions API set-up on a Woo store, with expected results at each step." |
| Handover note | "Write a one-page handover for this container: [paste summary]. Include owners, accounts and how to test." |
| Report summary | "Summarise these weekly numbers for a client in five lines, with one recommendation: [paste]. Do not invent numbers." |
| Offline conversions | "Design the data flow to send HubSpot deal stages to Meta and Google Ads as offline conversions. List the fields to save at the form." |
| Privacy review | "Review these event parameters for personal data that should not be sent to vendors: [list]." |
| Explain to a client | "Explain server-side tracking to a store owner in 150 words, no jargon, and one sentence on what it cannot do." |
Use it safely
- Do not paste access tokens, API keys, customer lists or personal data.
- Verify answers: open Preview, read the platform's documentation, test with a real order.
- Ask the assistant to say when it is unsure.
- Keep a record of what you changed and why.
Watch events arrive, live
A sample feed. In Open Tracked you see the same stream for your own store.
With an MCP server
If your assistant is connected to Tag Manager or to Open Tracked through MCP, it can read your actual containers and answer from them, instead of from what you paste. Even then, review its changes and publish yourself. See Open Tracked MCP server.
Runs on your domain, anywhere
Your tracking runs on your own domain, so a visit from any country is handled the same way.
Frequently asked questions
What are good prompts for tracking work?
Prompts that name the platform, state the goal, paste the evidence, give constraints and ask for a format such as a table or checklist.
Can I trust the answers?
Verify them in Preview and the platform's documentation. Assistants can be confidently wrong.
What should I not paste?
Access tokens, API keys, customer data and other personal data.
Can an assistant fix my container?
With the right connection it can propose or make changes in a workspace. Review and publish yourself.
Does Open Tracked have an MCP server?
Yes. See the MCP server guide for set-up.
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...
Guides · 5 min read
