Five analytics tools, five different user counts — one number everyone agreed on

The situation

A UK MarTech SaaS had accumulated 5 analytics/attribution tools over 3 years of growth: GA4, HubSpot, Segment, Mixpanel, and Meta Ads reporting. Every tool showed different user counts, conversion rates, and revenue attribution. The board asked for a single number to put in the investor deck. Nobody could agree.

The problem

Each tool had been set up independently with different user identity schemes: GA4 used anonymous client_ids, HubSpot used contact IDs, Segment was mapping user_id only after login, Mixpanel was using a separate distinct_id, and Meta was using fbp/fbc cookies. There was no shared identifier linking the same user across tools. Revenue was triple-attributed across Meta, Google, and "direct" depending on which tool you looked at.

What we did

1. Mapped user identity schema across all 5 tools — no shared identifier found
2. Implemented user_id as the canonical identifier: passed to GA4 user properties, Segment identify() calls, HubSpot contact ID linking, and Mixpanel alias
3. Built a stitching layer in BigQuery: joined all tool identifiers on email hash to create a unified user journey table
4. Built Looker Studio dashboard with Shopify/Stripe as revenue source of truth — each channel's contribution shown as % of verified revenue, not self-reported
5. Decommissioned Mixpanel overlap with GA4 — reduced tooling cost by £2,400/year

Why it worked

Attribution chaos is the inevitable result of adding tools faster than maintaining them. Each tool works in isolation — no tool natively talks to the others unless you build the stitching logic. The fix is a canonical user ID passed consistently to every tool from the moment of first identification, combined with a backend source-of-truth (Stripe/Shopify) that all channel claims are measured against.