Analytics & Reporting
GA4 setups that work, custom Looker dashboards, and attribution that survives Apple. Reporting that ends "but what does it mean?" meetings.
Three things go wrong, almost universally. One: GA4 is mis-implemented — events are noisy, conversions are double-counted, consent mode is wired up wrong. Two: the dashboard has sixty tabs nobody reads. Three: attribution stops working the moment a buyer touches an Apple device, which by now is most of your buyers. The result is a marketing team that ships work, ships reports, and never quite trusts what the numbers are saying. We fix all three.
GA4. Looker. Attribution. Cleanly.
GA4 from scratch
Event taxonomy, conversion definitions, consent mode v2, server-side container, BigQuery export. Everything documented.
Looker Studio suite
Weekly ops view, monthly executive view, quarterly board view. Each tuned to the audience.
Attribution modelling
Server-side UTM capture, deterministic identity stitching, modelled conversions for unobservable journeys. No pretending Apple ITP does not exist.
BigQuery layer
Daily fact tables for sessions, conversions, ad spend, and pipeline. Stable schemas your analysts can build on.
Monthly insight
Three-page written commentary explaining what moved, why, and what to change.
Documentation
Every event, every metric, every dashboard definition — written down. Your next analyst inherits a system, not a mystery.
Build. Stabilise. Hand over.
GA4 build
4–6 weeks. Event taxonomy, conversion definitions, consent mode, server-side, BigQuery export.
Dashboards
2–3 weeks once data is in place. Weekly, monthly, board.
Attribution
6–10 weeks depending on ad platforms and product surfaces in scope.
Documentation
Continuous. Every change captured. Your team owns the system before we leave.
Dashboard. Commentary. Backlog.
Templated 60-tab dashboards no one reads. Vanity-metric scorecards (sessions, bounce rate, pageviews). Attribution models that pretend to be deterministic when they cannot be. Reports that hide problems behind charts. If a number is bad, we say so on page one.
Questions, answered.
We already have GA4. Do you rebuild it?
Almost always, yes. Most GA4 implementations have at least 3–4 structural issues that make downstream reporting unreliable. We audit first; rebuild what needs rebuilding.
Do you work with platforms other than GA4?
Yes. We work with PostHog, Amplitude, Mixpanel, Heap, and bespoke event pipelines into Snowflake or BigQuery. GA4 is the default, not the only option.
Can you handle attribution for paid + organic together?
Yes. The full-funnel modelling layer combines paid platform data with server-side organic tracking and product analytics. That is the only attribution model worth trusting.
Who owns the dashboards after the engagement ends?
You do. Everything sits in your Google Cloud / Looker workspace. Every event, metric, and view is documented so your team can maintain it.