GA4 implementation & support

Google Analytics 4 that tells you the truth.

Most GA4 properties were migrated in a hurry and have been quietly wrong ever since. We audit what you've got, rebuild what needs rebuilding, and stay on hand when something breaks.

What it is

GA4 in plain English

Google Analytics 4 is the current version of Google Analytics, and it replaced Universal Analytics in July 2023. It isn't a new coat of paint on the old product — it's a different measurement model.

Universal Analytics thought in sessions and pageviews. GA4 thinks in events: every interaction, from a page view to a form submission to a purchase, is an event with parameters attached to it. That's more flexible and it's a much better fit for how people actually move between phones, laptops and apps.

The catch is that flexibility only pays off if someone decided, deliberately, what to collect and what to call it. GA4 will happily collect a pile of automatically-captured events and report on them, which is why so many properties look like they're working while telling you almost nothing useful.

Why yours probably isn't right

The July 2023 deadline meant a lot of migrations were done at speed, and many were Google's automated migration, which carried old goals across as conversions without redesigning anything underneath. That leaves you with a property shaped like Universal Analytics, running on an engine that doesn't work that way.

Add to that the things GA4 doesn't do unless you tell it to — filter your own staff's traffic, handle cross-domain journeys, mark the right events as key events, keep raw data longer than 14 months — and you get reports that are confidently, plausibly wrong.

Signs it needs attention

  • Conversion counts don't reconcile with your CRM or order data
  • A big chunk of traffic is sitting in (not set) or Direct
  • Event names are inconsistent — form_submit, formSubmit, Form Submitted
  • Your office or agency traffic is still counted in the numbers
  • Users jumping between your domains are counted twice
  • Nobody on the team opens GA4 unless they have to

How we help

What we actually do to your GA4

Audit it honestly

We go through data collection, event design, key events, attribution settings, filters, cross-domain configuration, data retention, audiences and integrations. You get a ranked list of what's wrong, what it's costing you in decisions, and what it takes to fix — not a 60-page PDF nobody reads.

Design the measurement, then build it

We start from your commercial questions and work backwards to the events and parameters that answer them. That becomes a measurement plan and a data layer specification your developers can implement from, with naming conventions that still make sense in two years.

Make conversions reconcile

Key events configured to match how your business actually counts a sale or a lead, deduplicated across form and thank-you-page triggers, with enhanced conversions where they help and a documented explanation of any gap that legitimately remains against your CRM.

Get the data out of GA4

The BigQuery export for unsampled, unlimited-retention raw data, and Looker Studio dashboards built on top of it that answer your team's real questions rather than showing every metric Google offers.

What you get

Deliverables

  • A written GA4 audit with issues ranked by business impact
  • A measurement plan tied to your commercial questions
  • A data layer specification your developers can build from
  • A configured GA4 property, QA'd in staging and production
  • BigQuery export configured, if it's worth it for you
  • Documentation and a training session for your team

Typical timings

1–2 wks
Audit, from kick-off to written findings
2–4 wks
Rebuild for a typical business or lead-gen site
4–8 wks
Ecommerce or multi-domain, including developer time

The variable is almost always how quickly your developers can implement the data layer. We'll flag that honestly at proposal stage rather than discovering it in week three.

Questions

GA4 questions we get a lot

Ours was set up during the migration. Why would it need work?

Because most migrations were done at speed against a hard deadline, and many were Google's automated migration, which translated old goals into conversions without redesigning anything. GA4 is an event-based model. A property that was translated rather than rebuilt tends to collect the wrong things, in the wrong shape, under the wrong names.

How long does it take?

An audit is one to two weeks. A rebuild for a typical business site is two to four weeks including QA. Ecommerce and multi-domain setups run four to eight, and most of that is waiting on developer time for the data layer rather than anything we're doing.

Do we need the BigQuery export?

If you care about data older than GA4's retention window, want analysis that isn't sampled, or need to join web data to CRM or order data — yes. The export is free from GA4's side and BigQuery storage costs most mid-sized sites a few pounds a month. If none of that applies to you, we'll say so.

Will we lose our historical data?

Restructuring how data is collected doesn't delete what's already there, but it does create a break in comparability — events collected under the old scheme won't line up neatly with the new one. We plan for that: annotate the change date, keep the old definitions documented, and where it matters, preserve a parallel view for a period so you can bridge the two.

Can you support it on an ongoing basis?

Yes. Retained support covers new tracking requests, re-testing after releases, keeping up with Google's platform changes, and investigating the "why has this number moved?" questions. Rolling monthly, cancellable with a month's notice.

Find out what your GA4 is getting wrong

Thirty minutes in your property, free. You'll get a straight answer on whether it's worth fixing.