What is a tracking plan, and how do you build one?
In short
A tracking plan is the reference document for a website's measurement: it lists every event you track, what triggers it, the parameters it sends, the conversion it feeds and the consent rule that applies. It is the developers' specification, the QA checklist before every release and the source of truth when numbers disagree. You build it from business goals, not from the tool. Without one, every redesign resets measurement to zero.
In this guide
What a tracking plan is for
- Getting everyone aligned. Marketing says what it wants to measure, developers know what to push, the implementer knows what to configure in Google Tag Manager.
- QA. Every row becomes a test: does the event fire at the right moment, with the right parameters, in the right consent state?
- Surviving redesigns. The site changes, the plan stays: you rebuild measurement from the plan, not from the old container.
- Settling discrepancies. When Google Ads, GA4 and your CRM disagree, the plan says what is supposed to be counted, and where.
What it contains
Each event takes one row, with the same columns.
| Column | What it specifies | Example |
|---|---|---|
| Event | The name sent to GA4 and ad platforms | generate_lead |
| Trigger | The exact moment it fires | Form submitted successfully, not the click on the button |
| Parameters | The information sent with it | form_id, lead_topic, value |
| Source | Where the data comes from | Pushed by the site into the dataLayer |
| Conversion | What it feeds in each platform | Primary action in Google Ads, key event in GA4 |
| Consent | The rule to respect | Analytics consent granted, advanced Consent Mode |
| QA | Test status | Passed on staging on launch day |
Around this table, a complete plan adds the dataLayer specification for developers, the custom dimensions to create in GA4, each platform's conversions, the consent rules, the QA tab and the version log.
How to build it, in six steps
- Start from business goals. List what the company needs to know: leads, quotes, sales, value. Every event in the plan should answer a business question.
- Audit what exists. Inventory the tags, triggers and conversions in place, and spot what no longer serves a purpose.
- Walk the site, journey by journey. Forms, quote or checkout funnel, appointment booking, downloads: every journey that leads to a conversion.
- Name the events. Use GA4's recommended events first (
generate_lead,qualify_lead,close_convert_lead,purchase): their predefined names and parameters feed reports directly. Stay within GA4's limits: 40 characters per event name, 25 parameters per event. - Specify the dataLayer. The site pushes business information into the data layer; Google Tag Manager reads it. The plan says, event by event, what to push and when.
- Test and version. Test every row on staging, then date the published version. Every change to the site goes through a new version of the plan.
Common mistakes
- Measuring the click instead of the success. A click on "Send" is not a lead: the form may fail or be rejected.
- Inconsistent names.
Lead,lead_formandcontact_formfor the same result: reports fragment. - Parameters never registered. A parameter sent to GA4 only shows up in reports once registered as a custom dimension, within a limit of 50 event-scoped dimensions on a standard property.
- Copying the old container during a redesign. You carry over dead tags and duplicates.
- Forgetting consent. A plan that does not say which event fires in which consent state cannot be tested.
- A plan that is never updated. Six months later, it no longer describes the site.
What we found at two clients
A van and heavy truck rental company, in the middle of a site redesign, had a GTM container that had grown without a plan: 53 of 73 triggers no longer served any purpose. The existing setup was audited, staging was explored page by page (305 pages, 25 quote journeys), then the tracking plan and the dataLayer specification were written before launch. The plan was approved before going live: 13 reliable events were kept, the rest rebuilt cleanly through the dataLayer.
A vehicle rental company in three markets, with a GTM container shared by about thirty brands, had business parameters that had been empty since the start of the year. The market-by-market review put a number on the gap: 34.5% of revenue without a transaction ID, or 158 purchases and €34,293 in one month in a single market. The fix plan was approved with the client and its developers.
Frequently asked questions
What is the difference between a tracking plan and a dataLayer?
The tracking plan is the document: it describes what you measure, when and why. The dataLayer, or data layer, is the technical object where the site pushes the information that Google Tag Manager then reads. The plan specifies what the dataLayer must contain; the dataLayer carries it out.
Does a small website need a tracking plan?
Yes, even a short one. Five or six rows are often enough: the conversions that matter, their triggers, their parameters and their consent rule. That is what lets you check that Google Ads learns from the right results, and avoid rebuilding everything at the next redesign.
What format should a tracking plan use?
A spreadsheet remains the most practical format: one row per event, one tab per topic (events, parameters, dataLayer, conversions, consent, QA, versions). At TAGBRUT, it is versioned and written so your team can implement it, with or without us.
Who should maintain the tracking plan?
A named person, on the company or provider side, who approves every change before it is published in Google Tag Manager. Every new page, form or ad platform goes through an update of the plan, then QA.
Is a tracking plan the same as a measurement plan?
Mostly. A measurement plan usually starts higher, from business goals and KPIs; a tracking plan goes down to each event, trigger and parameter. A good tracking plan includes both levels. In French, the term is "plan de marquage".
When to call in an expert
- You are preparing a redesign and want measurement to work on launch day.
- Nobody knows anymore what your Google Tag Manager container measures.
- Your developers are waiting for a dataLayer specification.
- Google Ads, GA4 and your CRM do not count the same thing.
The diagnosis is free: the gaps with a figure, their cause and the list of fixes, in writing within 2 business days of account access.
For a first check in two minutes, test your site with the tracking self-check, or read your consent signals with the gcs and gcd decoder.
Sources (4)
- Analytics Help, "[GA4] Recommended events": predefined names and parameters, lead generation events (`generate_lead`, `qualify_lead`, `close_convert_lead`) and ecommerce (`purchase`).
- Analytics Help, "Event collection limits": 40 characters per event name, 25 parameters per event.
- Analytics Help, "[GA4] About custom dimensions and metrics": 50 event-scoped dimensions on a standard property.
- Google for Developers, "The data layer": object used by Google Tag Manager and gtag.js to pass information to tags. developers.google.com/tag-platform/tag-manager/datalayer
More guides
- Consent Mode v2: basic or advanced, and how do you fix tracking after a cookie banner change?Read the guide
- How do you send CRM sales back to Google Ads?Read the guide
- How do you check that your Google Ads conversions are accurate?Read the guide
- Server-side tagging or Google Tag Gateway: which one do you need?Read the guide
- Consent Mode decoder: what do gcs and gcd mean?Open the tool