What should a measurement plan answer before a redesign?

A measurement plan is the agreement your team makes before anyone builds tracking: which outcomes matter, what exactly counts, where each one is measured, and who checks it. Without one, a redesign usually rebuilds the tracking the old site had, including its mistakes.

The questions a good plan answers

  1. Which business outcomes does the website support: sales, leads, signups, bookings?
  2. For each one, what exactly counts? A submitted form, an accepted lead, a paid order?
  3. Which event records it, with which details?
  4. Which platforms need to receive it: GA4, Google Ads, Meta, the CRM?
  5. How should it behave when a visitor accepts or declines tracking?
  6. Who builds it, and who confirms it works?
  7. How will you test it before launch and after?
  8. Which report shows it, and who reads that number?

Who should be in the room

A plan one person writes alone tends to miss the details only someone else knows. Get the people who build, receive and report on each number into one session before anyone opens Tag Manager.

  • Whoever owns the goal.Usually a marketing lead, who decides which outcomes matter.
  • Whoever runs the ad accounts.They know which conversions the bidding uses.
  • The web developer.They know how each form and checkout confirms success.
  • Whoever owns the CRM or order system.That record is what the other numbers get checked against.
  • Whoever answers for privacy.They give the consent column a real answer.

Bring the current site's list of forms and checkouts, last month's marketing report and the new site's page list.

A worked example

Measurement plan templateExample row, not from a client.
The template's columns, with the example row filled in
ColumnExample row
Business goalMore qualified sales conversations
Key actionDemo request
What countsA form submission the site accepts, not a button click
Eventform_success, with the form name
Sent toGA4 (key event), Google Ads (primary conversion), Meta (Lead)
ConsentAnalytics and ad platforms follow the visitor's choice
Built byWeb developer
Confirmed byAnalytics lead, with real test submissions
Reported inMonthly marketing dashboard, as Leads

Filling in the harder columns

Most columns fill themselves in. These six are where teams usually stall, so give them the most time in the session.

What counts
Write it so a salesperson would recognize it. If a form can fail validation or a payment can be declined, the click is an attempt. The confirmation or the order record is the outcome.
Event
Give each event one name and use it everywhere you can. List the details you'll want later, such as the form name or order value, and leave out anything personal like an email address.
Sent to
Mark the platform whose bidding uses this conversion. Test that one hardest, because a duplicate or a miss changes how your ad budget gets spent.
Consent
Write what should happen when a visitor accepts and when they decline. If the site uses Google Consent Mode, note it here.
Confirmed by
Name one person rather than a team, and the evidence they'll use: test submissions, a CRM record or a platform's diagnostics.
Reported in
Name the report and the label the number carries there. If the dashboard says Leads and GA4 says form_success, write both down.

Testing it before and after launch

A tag that shows up in Tag Manager's preview hasn't proven much yet. Work through these for every key action in the plan.

  1. Submit real test entries on the staging site, including one that should fail.
  2. Check that each event reaches GA4 with its details, in DebugView or the Realtime report.
  3. Decline tracking in the consent banner, repeat the test, and check what still fires.
  4. Check each ad platform's diagnostics or test events to see that the conversion arrived.
  5. Find the matching record in the CRM or order system.
  6. Repeat the same tests on the live site after launch. The live domain, banner and tag settings can differ from staging.
  7. For the first week, compare daily counts with the CRM or order system and look into any gap.

Where plans go wrong

  • Counting attempts instead of outcomes.
  • Agreeing on the plan but not on who tests it.
  • Skipping consent until after launch.
  • Writing it once and never updating it when the site changes.

Keeping the plan current

Give the plan one owner. Add a row whenever a form, checkout or tool changes, and read it through before each launch. When the next redesign comes, the plan is already written and only needs checking.

Related reading

Get the template

A Google Sheet with every column from the example above, plus a filled example row. Enter your email and I'll send you the link.

I'll email you the template and follow up once to see if it helped. No newsletter, no sequence. Privacy notice

Prefer to work through it together? I run measurement planning as a working session with your team.

A river bending past golden cottonwoods and red cliffs, with junipers in front.

Is a redesign on the calendar?

Tell me what is changing and which outcomes your team needs to keep counting.