Lorixen.
EN

Write a tracking plan before adding analytics events

Short answer: Before anyone adds another analytics event to your website or app, write a tracking plan: a short document listing the questions you want answered, each event that answers them, its exact name, the parameters it carries, where and when it fires, and who owns it. It prevents the usual mess of duplicate events, inconsistent names and reports nobody trusts. Prefer your analytics tool's built-in and recommended events where they fit, keep custom events few and well named, and test every event before it goes live.

Why analytics turns into a mess

Most businesses add tracking piece by piece. A developer adds "SignUp" for one campaign, a marketing agency adds "sign_up_complete" through a tag manager, the app team adds "registration" in the mobile app. A year later nobody knows which number is right, reports disagree, and decisions are made on gut feeling again.

A tracking plan is the cure: one agreed list of what is measured and how, written before implementation and kept up to date afterwards.

Start with questions, not events

List the business questions first. For example:

  • How many enquiries do we get, from which pages and campaigns?
  • Where do people drop out of the booking or checkout process?
  • Which features of the app do paying customers actually use?
  • How many people use the store finder, and do they then call or visit?

Each question suggests the events needed to answer it. If an event does not serve a question, leave it out. Fewer, cleaner events beat a hundred nobody reads.

Use what your tool already provides

Most analytics platforms collect some events automatically and recommend standard names for common actions. Google Analytics 4's documentation, for instance, describes automatically collected events, enhanced measurement events, recommended events, which you implement but which have predefined names and parameters and unlock existing and future reporting capabilities, and custom events that you define. It advises creating custom events only when no other event works for your use case, and notes that custom events do not show up in most standard reports, so you need custom reports for them.

So, for a sign-up or a purchase, use the tool's recommended event name rather than inventing your own.

What each entry in the plan contains

For every event, record:

  • Name, in a consistent style, for example lowercase with underscores: generate_lead, start_booking, complete_booking.
  • Description of the user action in plain words.
  • Trigger: exactly when it fires, such as "after the server confirms the booking", not "when the button is clicked".
  • Parameters: each with its name, type and example values, such as service_type, party_size or value.
  • Platforms: website, iOS app, Android app.
  • Owner: the person who approves changes.
  • Question it answers, linking back to your list.

Keep the plan in a shared document or, better, in a file in your code repository so changes are reviewed with the code.

Respect the tool's limits

Analytics platforms impose limits that a plan helps you stay within. GA4's collection limits page, for example, says it does not process an event name with more than 40 characters, allows 25 event parameters per event, and for app data streams allows 500 distinctly named events per app user. Other platforms have their own quotas. Short, reusable names and shared parameters keep you well inside them.

Agree how conversions are counted

Decide in the plan which events count as conversions or key events, and how each should be compared with your own records. A booking event in analytics will rarely match the booking system exactly, because of blocked scripts, consent choices and people using several devices, so state the expected gap and check it monthly. If analytics suddenly drifts much further from your records, that is usually a broken event, not a change in customer behaviour, and the plan tells you exactly where to look.

Privacy and data hygiene

Never send personal information, such as names, email addresses or phone numbers, as event parameters. Use internal IDs where you need to link events to records, and check your consent setup controls what is sent. Note in the plan which events are affected by consent choices.

Implementing and testing

  1. Implement events exactly as named in the plan, ideally through one shared helper in code or a single tag manager container.
  2. Test in the tool's debug or real-time view on staging: every event fires once, at the right moment, with the right parameters.
  3. Check edge cases: back-button navigation, page refreshes, failed payments and app restarts often cause double or missing events.
  4. Mark each event in the plan as "planned", "implemented" or "verified".

Keeping the plan alive

  • Any new event requires a plan update first.
  • Remove events nobody uses; they clutter reports.
  • Review the plan when you redesign key pages or launch new features.
  • Version the plan, so you can explain why numbers changed on a certain date.

A worked example

A hypothetical Halifax physiotherapy clinic group ran a website, an online booking tool and an app. Its analytics had four different events for a completed booking, one firing on button click even when payment failed, and nobody could say how many bookings came from the website.

The clinic's developer wrote a tracking plan around five questions, kept recommended event names where they existed, defined a single complete_booking event fired only after confirmation with service_type and clinic_location parameters, and retired the duplicates. Each event was verified in debug mode on staging. Monthly booking figures in analytics now match the booking system closely, and the marketing team can compare locations with confidence.

FAQ

Do small sites need a tracking plan?

Yes, even a one-page plan. The smaller the team, the easier it is to keep consistent from the start.

Who should write it?

Someone who understands the business questions, working with whoever implements the tracking.

What about server-side events?

Include them in the same plan, noting where they are sent from, so web, app and server events use the same names.

If your analytics events have grown messy and reports no longer agree, send us a list of what you track today.

Sources

WhatsAppBook a call