Case study · Automotive
How we restored +37% of ad-platform signal for a UK marketplace.
A 405-day study of organic and paid uplift.
+37%
Recovered organic sessions, 405-day study
The challenge
What wasn’t working
A high-traffic UK marketplace was watching paid performance degrade quarter after quarter. The ad platforms were reporting fewer conversions than the backend, but nobody could explain the gap with confidence.
Client-side tags were being blocked by browsers and consent tools, deduplication between web and backend events was inconsistent, and the warehouse was missing the join keys needed to validate any of it. Bid algorithms were starving and the marketing team had no auditable answer for the finance director.
What we did
How we delivered it
01 Audit and gap analysis
Two weeks mapping every tag, event and platform integration. Signal loss measured in exact percentages between browser, backend and ad platforms.
02 Server-side container build
Stood up a server-side tagging container on a custom subdomain, ported the priority conversion events, added deduplication keys.
03 Consent and identifier hygiene
Reconciled consent state across CMP, GTM and server. Backfilled hashed identifiers where the user had consented to advertising.
04 BigQuery as source of truth
Streamed deduplicated events into BigQuery, joined to CRM stages and revenue. Built audit dashboards for marketing and finance.
05 Platform reconciliation
Wired the priority events into Google Ads and Meta using enhanced server-side feeds. Validated conversion lift channel by channel.
06 Ongoing health monitoring
Anomaly alerts on event volumes and a weekly QA cycle so regressions are caught before media buyers notice.
The result
The outcomes
+37%
Recovered organic sessions, sustained over a 405-day study window
+42%
More conversion signal returned to ad platforms versus the previous client-side setup
6 weeks
From kickoff to first production stream into BigQuery
100%
Of priority paid events deduplicated and consent-aware end to end
“The team treated our measurement like it was their own product. We finally have numbers we trust to optimise on.”
Head of Performance, automotive marketplace
What to read next