tracking infrastructure4 min readBy Phloz team

The measurement plan every agency should build per client

Most agencies 'set up GA4 and hope.' A measurement plan is the one-page artifact that maps a client's goals to the exact events you track — built before you touch a tag. The template and how to use it.

TL;DR

A measurement plan is a one-page document that maps a client's business goals → the user actions that signal them → the exact events/conversions you'll track → where each is configured → who verifies it. Built before you touch a tag, it's what separates "we set up GA4 and hoped" from tracking that actually answers the client's questions. It prevents the two classic failures: tracking the wrong things (vanity events nobody acts on) and missing the right ones (the conversion the client's budget actually depends on). It also becomes the contract you verify against — every audit, every QA pass, every report traces back to it. Below: the template, how to fill it per client, and how to keep it alive.


Walk into a typical client's GA4 and you'll find tracking that grew by accretion — a purchase here, three test conversions from 2023 there, half the marked conversions misfiring, and no record of why any of it exists or whether it covers what the client actually cares about. That's the cost of skipping the measurement plan. It's the cheapest, highest-leverage hour in the whole engagement, and almost nobody does it.

What a measurement plan is (and isn't)

It is a short artifact that answers, for one client: what does success look like, what actions signal it, what will we track to see those actions, where does each live, and who confirms it works?

It is not a tagging spec, a GA4 export, or a 40-tab spreadsheet. One page (or one screen) per client. If it's longer than that, it's not a plan — it's a project nobody will maintain.

The template

For each client, fill five columns:

Business goalSignal (user action)Event / conversionConfigured inOwner / verified
Generate qualified leadsSubmits the contact formgenerate_lead (key event)GTM trigger → GA4; marked conversion in GA4Verified in DebugView ✓
Drive demo bookingsCompletes Calendly bookingbook_demo (key event)GTM + cross-domain to CalendlyPending QA
Ecommerce revenueCompletes checkoutpurchase w/ value+currency+transaction_idGTM → GA4 + Ads + Meta CAPIVerified ✓
Phone leadsClicks call / callscontact (offline import)Call tracking → offline conversionOwner: PPC lead

That's the whole tool. The discipline is in filling it in this order — goal first, event last — so you only track what maps to something the client cares about.

How to build one (per client, ~1 hour)

  1. Start with the client's goals, in their words. "More qualified leads," "more demo bookings," "more revenue." Not metrics yet — outcomes.
  2. Translate each goal to a user action that reliably signals it. "Qualified lead" → submits the specific form (not any form). Be precise; this is where vanity metrics get filtered out.
  3. Name the event + its parameters. What event fires, and what data it must carry — a revenue event without value/currency is worse than useless. Decide the key parameters now.
  4. Decide where each is configured. Browser event via GTM? Automatic via GA4 Enhanced Measurement? Server-side? Marked as a conversion in GA4 and mapped into Ads/Meta? (If "GTM or GA4?" trips you up, here's the split.)
  5. Assign an owner + verification. Who confirms each event actually fires correctly, and when. An unverified plan is a wish.

Why it pays off all engagement long

The plan isn't a one-time deliverable you file and forget — it's the reference everything else checks against:

  • Build: you implement to the plan, so nothing important gets missed and nothing pointless gets added.
  • QA: the pre-launch verification pass is just "does each row of the plan actually fire correctly?"
  • Audit: a GA4 audit becomes "does live tracking still match the plan?" — misfires and orphaned conversions stand out because the plan says what should exist.
  • Reporting: every number in the client report traces to a row, so "where does this come from?" always has an answer.

It's also the artifact that makes a week-one tracking audit on a new client productive — you're auditing against an intended state, not guessing what the previous agency meant. (And if you need a starting list of what's usually worth tracking, 21 things to track per client is a good prompt — but the plan is what makes it this client's list, not a generic one.)

Keep it alive

A measurement plan rots the moment the client's goals change and the plan doesn't. Revisit it when the client adds a product line, a new funnel, a new channel — and keep it somewhere attached to the client, not in a folder nobody opens. That's exactly what Phloz is for: the plan lives on the client record next to the tracking-infrastructure map that implements it, so the intended state and the live state sit side by side and drift is visible. The CRM for SEO agencies and pricing pages cover the workflow — but the template above costs an hour and a doc, and it's the hour that makes every other tracking hour count.