Replacing the lossy spreadsheet with a direct Data Manager API integration: CRM Quote and Order conversions pushed back to Google Ads and GA4, tied to the original ad click — and when a record doesn't make it, you see exactly which, and why.
RFQs are the conversion every downstream decision depends on. The job is to count them accurately, attribute them to the right source, and make the whole pipeline transparent — first as a tool your team reads, then as an agent that runs itself and escalates only when a human should decide.
Offline conversions into Google Ads + events in GA4, with per-row visibility into what didn't import — orchestrated in n8n.
Regularly reconcile across CRM, GA4, Google Ads, and Microsoft — website RFQs only, excluding unsettled days, explaining the divergence rather than treating it as a bug.
A tool your team reads. Live build.
An agent you supervise by exception.
The working artifacts behind the engagement — each viewable here, kept current as the build progresses.
The full product mockup — overview, conversions, match rate, reconciliation, runs, and the Phase 2 agent.
Current state, the 8 gaps where attribution leaks, and the target-state design.
gclid + Enhanced Conversions, measured on live data: +85% quotes, +152% orders.
Which converted RFQs we can attribute, and why the rest can't — the honest read.
The live Supabase schema, table by table — the source of truth, documented.
Objective 1, phased — what's built, what's gated, and the sequence to go-live.
