← Back to case studies
BigQuery Looker Studio GA4 Attribution

FinTech SaaS — Attribution Across 6 Paid Channels

B2C financial product · performance marketing team of 8 · $800k monthly ad spend

22% Ad spend waste eliminated within 60 days
Problem

Six platforms, six different conversion numbers — no one trusted the data

The performance team was running paid campaigns across Google, Meta, TikTok, Snapchat, LinkedIn, and Apple Search Ads. Each platform reported its own conversion count. None of them agreed.

Last-click attribution in Google Analytics assigned 80%+ of credit to Google Ads — the last touchpoint before conversion. Every other channel looked like it wasn't working. Budget decisions were made on this basis: TikTok spend was repeatedly cut despite the channel playing a documented role at the top of funnel.

No one could answer the question: which channels are actually driving revenue, and in what proportion? Without that answer, every budget reallocation was a guess.

Approach

Unified attribution pipeline into BigQuery with data-driven modeling

Built a data pipeline that pulled raw conversion data from all six platform APIs into a single BigQuery dataset. GA4's BigQuery export served as the source of truth for on-site behavior — session data, event sequences, and user journeys.

Implemented a data-driven attribution model using Markov chain analysis on the BigQuery data. Each channel received credit proportional to its actual contribution to conversions — measured by its counterfactual impact on the conversion path.

Built a Looker Studio dashboard showing channel contribution by funnel stage: awareness, consideration, and conversion. For the first time, the team could see how TikTok drove top-of-funnel sessions that converted weeks later through Google.

Result
22% waste cut · TikTok vindicated

Budget was reallocated within 60 days based on the new attribution data. Channels that appeared underperforming under last-click (TikTok, Snapchat) were shown to drive significant upper-funnel contribution — their cuts were reversed.

Channels that appeared high-performing under last-click (branded search) received proportionally less credit once the full path was modeled. The result was a more evenly distributed budget with better actual performance.

TikTok's true upper-funnel contribution was quantified and documented for the first time — a finding that changed how the team structured creative and spend for that channel going forward.

What was delivered
Multi-platform API data pipeline into BigQuery (6 platforms)
GA4 BigQuery export configuration and schema alignment
Markov chain attribution model built in BigQuery SQL
Looker Studio dashboard: channel contribution by funnel stage
Automated daily refresh for all platform data
Documentation and training session for the performance team
Tools BigQuery GA4 Looker Studio Supermetrics Python

Running across multiple paid channels with conflicting data? Let's fix the attribution.

Book a free audit call