How I rebuild broken analytics.
Most analytics rebuilds aren't a tooling problem — they're a documentation problem. The fastest path back to trust is writing down what should be measured, then making the tags reflect the doc.
Three dashboards.
Three answers.
A migration was done in a sprint. Tags fire twice. Conversions count on the wrong step. The marketing dashboard, the finance dashboard, and the GA4 explore all give different numbers — and nobody knows which one is right.
The team stops shipping decisions off data, because every decision needs a 20-minute "can we double-check this number?" conversation. That's the symptom I get called for.
Rebuild the data layer.
Then the tags.
I start with the spec, not the tool. Every event the business needs, named in plain English. Every parameter typed. Every conversion mapped to a single defined step. Marketing and finance review the spec *together* before a single tag changes — the goal is one document everyone signs off on.
Then the GTM rebuild — versioned containers, paired QA, a documented publish cadence. Three weeks in there's a clean GA4 property, documented events, and a dashboard the team agrees to.
The iterations
- Week 1. Audit existing tracking. Catalog every gap, doubled fire, and broken conversion. Write the data layer spec the team wishes existed.
- Week 2. Rebuild GTM container against the spec. Paired QA between me and a stakeholder. No publishing yet — staging only.
- Week 3 — 4. Publish in waves. Watch live data against expected. Hand off the spec, the container, and a maintenance doc.
Tracking that holds.
Without me.
Three artifacts ship with the engagement: a written data-layer spec, a versioned GTM container, and a maintenance doc your team can use after I leave. Plus a handoff session walking through what changed and why.
The point is not that I built it. The point is that the team can edit it.