Home  /  Services  /  Measurement & Tracking

07 · Measurement & Tracking · Fixed scope

The ad platforms report your conversions. Have you ever counted them yourself?

Google says 40 conversions this month. Meta says 35. The CRM received 50 leads, and the two reports add up to 75. The thank-you page counts a conversion every time someone refreshes it, and half the tags quietly died when the site was rebuilt last spring. Budget is allocated on those numbers every month.

This project rebuilds measurement from the definition up: a written plan for what to count, GA4 and Google Tag Manager in your own accounts, the five conversions that actually make money, ad platform connections, consent handling, and a QA report in which every event was fired and checked by hand, plus a dashboard. Handed over in about four weeks — and from then on, any agency that comes after us sees the same numbers.

Who this is for

Three situations we see most.

Online store

Ads optimise on "conversions", and each one is counted twice.

The purchase event fires in GA4 and in the ad pixel separately, sometimes on every refresh. The algorithm learns from an inflated signal, ROAS looks better than it is, and budget flows to the wrong campaigns.

B2B

Enquiries arrive. Nobody knows where from.

Form submissions pile up in the inbox, but GA4 has no event for them, and no lead is tied to a campaign or a keyword. Whether to raise or cut the marketing budget is decided by feel.

Any company

The site or the agency changed, and nobody knows what the tags do.

Tag Manager holds sixty unnamed tags and the person who built them has left. GA4 keeps data for two months by default, so while the check is postponed, the history to compare against disappears too.

Not for you ifYou have an analytics owner in-house and your conversion events were verified within the last quarter, or the site has too little traffic yet to count anything. Both are in the honest limits below, and we will say so in the audit.

Why it matters

When measurement is wrong, every decision above it is wrong.

Each platform grades its own homework

Google Ads and Meta each attribute the same purchase to themselves. Add the two reports and the total exceeds actual revenue. Without a number counted in your account by your rules, it is a match with no referee.

A broken tag makes no sound

Change one button and the trigger that captured its click simply stops. No error, no warning. Conversions appear to fall, and someone switches off the campaign.

Data collected without consent cannot be used

Since March 2024, Google Ads conversions sent from EU visitors without Consent Mode are not used for ad optimisation. Most sites have a banner installed and tags that ignore it.

The past cannot be recovered

GA4 does not backfill. A conversion not captured today is a conversion that never existed. The best time to fix measurement is before scaling the campaigns — which is now.

What you get

Eight parts, all in your accounts.

01 · Measurement plan

KPIs derived from business goals, the events to count with naming rules, each event's trigger condition and parameters, and a dataLayer specification. One document a developer or the next agency can follow as written.

02 · GA4 property

Data streams, internal traffic excluded, data retention raised to 14 months, cross-domain and referral exclusions, key events, and links to Google Ads and Search Console. An existing property is cleaned up; if there is none, one is built.

03 · Google Tag Manager container

Tags, triggers and variables under a naming convention, a note on every version, dead tags removed. Scripts hard-coded into the site move into the container, so future changes need no code deployment.

04 · Five key conversions

Form submission, phone and email clicks, purchase or signup, booking, download — the five that fit your business. Deduplicated, with conversion values where possible. No thank-you-page counting that fires again on refresh.

05 · Ad platform connections

Google Ads conversion import, Meta Pixel managed from the container, enhanced conversions. When Korea is in scope, Naver Search Ads conversion scripts and Kakao Pixel in the same container.

06 · Consent Mode v2

The cookie banner actually connected to the tags. Before consent the tags wait; on refusal they send consent signals only. If you have EU or UK visitors, you lose ad optimisation data without this.

07 · Live-verified QA report

Every event fired in a real browser and confirmed in GTM preview, GA4 DebugView and the ad platforms' diagnostics. A table of what was verified and how is part of the handover.

08 · Dashboard and handover

One Looker Studio page — conversions by channel, cost per conversion, top pages. A working session with the owner, and a checklist to follow when adding tags later.

What lands in the folderThe measurement plan, the GTM container under your ownership with its naming convention applied, the configured GA4 property, the QA report, the dashboard, and a GTM container export file. Nothing sits in an account of ours.

How it is built

Design first. Tags after.

01

Decide what to count before counting

Most measurement failures are failures of definition, not technology. Until "a lead" is defined as a form submission, a valid form submission, or a sales-accepted lead, no tag can produce the right number. The plan comes before the tags, and the build starts after you approve it.

  • Five events, not more — only what decisions are made on, not everything that can be counted. The rest is noise.
  • A naming convention — so a new owner six months from now can tell what a tag does from its name.
  • The plan is yours — the next agency counts by the same definitions.
02

In your accounts, under your name

The GA4 property and the GTM container are created under your Google account, and we hold edit access only for the duration of the work. A container built in an agency's account becomes a hostage when the contract ends.

  • Minimum access — when the work is done, we are the ones who tell you to downgrade our permissions.
  • A note on every version — every GTM publish records what changed and why.
  • Hard-coding removed — scripts scattered through the site's code gathered into one container.
03

Verified by actually clicking

A tag being "published" and an event being "captured" are different things. Every event is fired in a real browser and confirmed three times — in GTM preview, in GA4 DebugView, and in the ad platform's diagnostics. That record is the QA report.

  • Duplicate check — counted once across refresh, back button and resubmission.
  • Device check — the same event captured in mobile Safari.
  • Consent check — tags actually stop when consent is declined.
04

The dashboard is one page

Nobody reads a thirty-page report. Conversions by channel, cost per conversion and the top-performing pages sit on one screen, and those numbers come from your GA4, not from the ad platforms. Whether we write your fortnightly report or someone else does, it is the same source.

  • Side by side with platform reports — what Google claims and what you counted, in the same table.
  • A free tool — Looker Studio costs nothing and stays in your account.
  • Extensible — if this leads to a retainer, this dashboard is the base of the fortnightly report.
05

Korea, if it is in scope

Naver and Kakao do not read Google tags. Naver Search Ads conversion scripts, Kakao Pixel and Naver Analytics are managed in the same GTM container so Korean channel conversions land on the same dashboard.

  • Same container, same rules — Korean channels do not become a separate pile of scripts.
  • Ad accounts are separate — the Naver and Kakao ad accounts themselves need a Korean business registration, so they are scoped in our Korea market entry work.

Process

Four steps, about four weeks.

Week 1

Audit and plan

Read access to GA4, GTM and the ad accounts to see what is captured now and what is counted twice. KPIs agreed, the measurement plan written and approved.

Week 2

Build

GA4 configured, the GTM container rebuilt, five conversions instrumented, ad platforms connected, Consent Mode wired. On a custom-built site, dataLayer snippets go to your developer.

Week 3

Verify

Every event fired in a real browser and confirmed in GTM preview, GA4 DebugView and the ad platforms' diagnostics. Duplicate, device and consent checks, then the QA report.

Week 4

Dashboard and handover

Looker Studio dashboard built, a working session with the owner, a checklist for adding tags later. We downgrade our own access and finish.

What we measureWhether all five conversions are still being captured thirty days after handover, and whether GA4's conversion count matches the real count in your CRM or order system within ten percent. Session counts and event totals are not reported.

Honest limits

What this project will not do.

01

It will not recover past data. GA4 does not backfill. Numbers are accurate from the handover date onward; comparisons with earlier periods are indicative only.

02

Server-side tagging and the Meta Conversions API are outside the base scope. They matter in industries where browser blocking is heavy, but they add a monthly paid infrastructure at your cost and a different level of operation. Scoped separately when needed.

03

No apps. This covers website measurement only. Firebase instrumentation for iOS and Android apps is a different job.

04

Measurement does not create demand. Accurate numbers change decisions; they do not make customers. Under a few hundred visits a month there is little to count, and that money is better spent creating visits first.

05

If you have an analytics owner in-house, you may not need us. A company with a record of verifying its conversion events last quarter can do most of this internally. The case for this project is a company where the person who set it up has left and nobody trusts the numbers now.

FAQ

Measurement and tracking, answered.

Which website platforms can you work on?+
Any platform that can carry a Google Tag Manager container — Shopify, WordPress and WooCommerce, Webflow, Wix and others. Custom-built sites work too, but capturing purchases and form submissions there needs your developer to install the dataLayer snippets we write. The snippets are in the measurement plan verbatim, so the developer's part is usually under a day.
We already have GA4. Why do we need Google Tag Manager?+
GA4 counts; Tag Manager decides what to count and when. With GA4 alone you get page views and automatic events, but conversions — form submissions, purchases, phone clicks — are mostly missed or require scripts hard-coded into the site. Kept in a container, later changes need no code deployment, and ad pixels are managed in the same place.
What do you need from us to start?+
Edit access to your Google accounts for GA4, Tag Manager and Ads, access to the Meta ad account, a login to your CMS or a developer contact, the login for your cookie banner tool if you have one, and one person who can decide what counts as a conversion. On a custom-built site, about a day of developer time in week two.
How long does it take?+
About four weeks from access: audit and measurement plan in week one, build in week two, verification in week three, dashboard and handover in week four. If developer cooperation is slow it stretches by that much, and we say so in advance.
Do you do server-side tagging or the Meta Conversions API?+
Not in the base scope. Both help in industries hit hard by ad blocking and iOS restrictions, but they add a monthly paid infrastructure such as Stape at your cost, and set up wrongly they double-count events. The right order is to make browser measurement accurate first, then scope server-side separately if it is needed.
What about GDPR and cookie consent?+
We set up Consent Mode v2 so tags actually wait or stop according to the banner's consent state. The banner tool itself — Cookiebot, OneTrust and the like — is chosen and paid for by you; we do the connection and the verification. We are not lawyers, so which banner wording is sufficient in which region needs legal review.
Are Korean channels covered?+
When Korea is in scope, Naver Search Ads conversion scripts, Kakao Pixel and Naver Analytics are managed in the same Tag Manager container so Korean channel conversions land on the same dashboard. The Naver and Kakao ad accounts themselves require a Korean business registration, so account setup is handled in our Korea market entry work.

Also

Paid Advertising →The retainer that optimises on the conversions captured here. We do not start it without measurement. Web & Landing Pages →Page builds and CRO whose tests are judged on the same events. Korea Market Entry →Naver and Kakao ad accounts, and Korean channel conversions on the same dashboard.

If you have read this far, the website audit is free.

The free audit is a website audit: how you show up in search and in AI search, whether basic tracking is alive, and the first three things we would fix, in writing. It is the same audit whichever page you came from, and it costs nothing. If it is Measurement & Tracking itself you want, say so in the same form and we will answer about that instead.

Get a free website audit

Or ask about Measurement & Tracking →