How an eCommerce Company Drove $14M in Contribution Margin in 9 Months

The gap between revenue growth and margin growth

This U.S. fashion brand operates across multiple retail locations with an $50–$500 transaction range, serving both new and returning customers across online and offline channels. Revenue was growing — but the team had no way to see whether that growth was generating margin. Campaign-level reporting showed ROAS; it did not show fully loaded variable contribution margin (FLM). Budget allocation was effectively blind to profitability.

Three problems compounded the challenge: there was no reliable way to track performance across all channels in real time using both last-click and multi-touch attribution; the investment mix between new and existing buyers had never been separated and modeled; and the data had enough operational inconsistencies that the team had lost confidence in the numbers they did have.

Building the measurement infrastructure

We built the first-ever Marketing Mix Modeling system for this brand — running 3,000 model iterations to reach 94% accuracy, incorporating category-level data that had never been included in previous analysis. Alongside MMM, we deployed a live Multi-Touch Attribution model to provide more precise budget allocation signals, particularly for top-of-funnel investment where last-click attribution systematically undervalued brand-building channels.

We also implemented a differentiated strategy for new vs. existing buyers — building separate planning and reporting views so the team could see, for the first time, which activity was driving new customer acquisition and which was driving repeat purchase. Real-time tracking across platforms, channels, and campaigns enabled day-to-day FLM-based spending adjustments.

$13.9M in margin from the same investment

By optimizing media investment against fully loaded margin rather than ROAS, the team reallocated budget toward the channels and cohorts producing genuine contribution — not just revenue. Over 9 months, this discipline delivered a $13.9M year-over-year increase in fully loaded variable contribution margin. The model continues to run, making the Capital Allocation Loop more precise with each cycle.