Agency specialized inGoogle Analytics 4with tailored implementation, data modeling and advanced exploitation.
We set up Google Analytics 4 with senior judgment: consistent event taxonomy, useful dimensions, server-side measurement where it adds value, correct consent management, and data you can defend before finance and before the ICO / AEPD / CNIL.
GA4 is not a "turn on and done" button. It’s a platform that gives you what you ask of it — and by default you don’t ask much. The difference between an install that helps you decide and one that just fills the monthly report is the ten hours of instrumentation almost nobody does.
Typical dimensions of a senior implementation
Not every one is always included. We design for real business goals.
Conversion events
Defined by real business objective, not defaults. With parameters that allow segmentation (channel, campaign, category, value).
Custom dimensions
User and session data that enrich reports: customer type, plan, segment, geography properly resolved.
Session quality metrics
Time engaged, scroll depth, meaningful interactions. To separate real traffic from bounces.
Enhanced ecommerce
Full ecommerce data layer implementation when applicable: view_item, add_to_cart, checkout, purchase with correct item-scope.
Cross-domain / cross-device
Correct setup when the user navigates between owned domains (blog → store → app) or between devices.
CRM/ERP reconciliation
Match between what GA4 reports and what appears in CRM/ERP. Documentation of expected gaps — because they always exist.
How we set up GA4 when it matters
From capture to report, with controls at every layer.
- L01
Capture
Client-side (gtag / GTM) for the basics + server-side (GTM SS, Cloudflare Workers) for critical conversions resistant to blockers.
- L02
Enrichment
User properties and session properties pushed from the site itself or from CRM. GA4 moves from "what they do" to "what they do and who they are".
- L03
Consent
CMP integrated with Consent Mode v2. Configuration of what’s sent without consent (predictive modeling) and what requires opt-in.
- L04
Validation
Testing with GA Debugger, Tag Assistant, GA4 DebugView. No event goes to production without verification.
- L05
Reporting
Custom explorations + BigQuery integration when volume or analytical need calls for it. Dashboards in Looker Studio, Metabase or similar.
Tangible deliverables
Concrete pieces ready for operations.
Measurement plan
Document with business goals → KPIs → events → parameters. The source of truth for instrumentation.
Documented configuration
GA4 setup with every event, dimension and filter documented. For internal audit or handover to a new team.
Explorations & reports
Collection of custom views in GA4 for the recurring business questions. No need to reinvent them every month.
BigQuery integration
Native export to BigQuery when applicable, with SQL views for the questions GA4 doesn’t answer well.
Maintenance playbook
How to keep instrumentation healthy: what to check, how often, which signals indicate drift.
Team training
Handover sessions for marketing and BI. No one has to "wait for the data person" to answer a basic question.
GA4 governance principles
Installing GA4 is 30%. The other two thirds are keeping it healthy.
- 01
Written naming convention
Every event and parameter follows a documented convention. No "boton_click" mixed with "buttonClick" and "btn_press".
- 02
Ownership per event
Every event has an owner in the business. Without this, "orphan" events become noise over time.
- 03
Continuous testing
Changes on the site can break events. Automatic monitoring with alerts when a critical event stops flowing.
- 04
Quarterly review
Every quarter, pruning of events no longer used and additions of ones now needed. Instrumentation is cleaned — not accumulated.
- 05
Active compliance
GDPR and ePrivacy are not "install a CMP and forget". We periodically review what’s sent, with what consent, and from where.
Honest questions before hiring
Can you audit the GA4 we already have before rebuilding?
It’s the most common approach. Often the problem isn’t scrap and start over — it’s identifying what’s fine, what needs fixing and what’s missing. We’ll deliver a report with prioritized findings before proposing any large project.
Do we have to move everything to server-side?
No. Server-side adds value on critical conversions, traffic blocked by ITP/adblockers and fine control over what’s sent. For many events, client-side is still enough. We design hybrid — without forcing a full rebuild.
How do you address "the numbers don’t match the CRM"?
By documenting expected gaps. They will never match 100% (different attribution windows, different attribution, offline vs online). What we can do is document how much difference is "normal" and which differences are a signal of a real problem.
Do you work with GTM (client + server) or only direct implementation?
Both. GTM has maintenance advantages and server-side GTM offers very fine control. Direct implementation (gtag) can make sense for very performance-sensitive sites. We choose per project — not by habit.
And Google Analytics Universal / historical GA3?
Google stopped collecting data back in 2023-2024. If you still need historical access, we help preserve it (export to BigQuery, local snapshot) and map old → new metrics for historical comparisons.
Does GA4 give you the data you need, or only what it knows how to give?
Tell us which business questions aren’t answered today by your analytics. We’ll see if it’s a configuration issue, an architecture issue, or something else.
Let’s talk?GA4 audit under NDA. No generic proposals.