The Challenge
A fast-growing direct-to-consumer skincare brand focused its entire media mix on paid search, paid social, and influencer whitelisting. While each platform reported healthy returns, their combined claims added up to 160% of actual orders recorded by their Shopify backend.
Platform reported numbers saw branded search and retargeting taking most of the credit and, with it, most of the budget, making the upper-funnel prospecting campaigns filling the funnel in the first place look expensive in every report. With only 18 months of advertising history, no established seasonality and a digital-only channel mix, marketing mix modelling was an option the data could not yet support.
The business needed a measurement approach suited to the current business reality, providing a single view of channel contribution to fuel their growth until MMM became viable.
The Solution
An audit of the client’s server-side tracking confirmed clickstream events, UTM taxonomy, and Shopify conversion records were consistent and linkable across the brand’s typical two-week path to purchase.
Stitching those journeys together produced a complete picture of customer behaviour, allowing a data-driven attribution model to distribute credit based on actual contribution. Removing any single channel from the observed journeys showed exactly how many conversions depended on it, bypassing the blind spots of platform-specific attribution, which simply cannot see the full journey.
The results fed a share of spend versus share of effect view that the founders reviewed weekly, with suggested reallocations validated via holdout audiences before any budgets were moved.
The Impact
Prospecting on paid social turned out to be earning nearly twice the credit last click had assigned it, while branded search was largely harvesting demand that other channels had created. Shifting budget towards these under-credited channels cut blended CAC by 18% on the same overall budget.
New customer growth followed the reallocation, with first-time orders up 14% over the following two quarters as prospecting budgets stopped competing with retargeting for the same credit.
The weekly attribution review has replaced platform reporting as the reference point in budget conversations, with the underlying data pipelines created for the model acting as groundwork for a future MMM.
Key Takeaways
- Right-Sized Measurement - Provided a clear channel view for a digital-first brand before enough historical data accrued for an MMM.
- Tracking Before Modelling – Verified server-side events and UTM taxonomies to ensure multi-touch journey stitching was completely reliable.
- Data-Driven Attribution - Distributed conversion credit based on observed journey dependencies, exposing heavily over-credited lower-funnel lines.
- De-risked Reallocation – Moved live marketing budget only after holdout audience tests confirmed the model’s performance read.
- Commercial Results – Reduced blended CAC by 18% and expanded first-time customer orders by 14% within an unchanged total budget.
Tools and Techniques
- Multi touch attribution
- Tracking and UTM taxonomy audit - profiling server-side tracking health and validating logging parameters
- Data-driven attribution (Markov Chain) - calculating channel removal effects to isolate true incremental contribution
- BigQuery - data extraction and central storage for clickstream and advertising data
- R - data handling and algorithmic attribution modelling
- Shopify - source extraction for conversion and customer data records
- Looker Studio - share of spend and share of effect performance reporting
Related services
No matching items