Case study · Sport equipment
Three markets read together, not one at a time.
A European maker of technical equipment for high-performance sport, modelled across France, Spain and Japan. The challenge is not reading one market: it is reading three at once, and seeing where they behave alike and where each follows its own dynamics.
The project
Sector
Technical equipment for high-performance sport.
Markets
France, Spain and Japan, modelled together.
AD cube module
Marketing Mix Model.
Sales channels
Online and in store, direct-to-customer and wholesale.
The decision challenge
What actually generates revenue, across three different markets.
The underlying question is about revenue: what generates it, and how the mix of levers behaves in each of the three countries. Compared with a single-market project the complexity is higher, because reading the overall result is not enough: the brand needs to see where France, Spain and Japan move alike and where country-specific differences appear.
There is a second element, tied to how the brand sells. It has full visibility of direct-to-customer sales, which remain the focus of the analysis. Of sales into the retail channel it knows only sell-in precisely, what it sells to dealers and retailers, but not sell-out, what those then resell to the end customer. The analysis therefore compares two parallel readings, the sell-in trend and the direct-to-consumer sell-out trend, to see how closely the two move together and to choose the more consolidated trend to build the model on.
The method
Revenue broken into four components.
The model reads the brand’s revenue as the sum of four parts, each isolated from the others.
Baseline
Organic sales: brand strength built over time, long-term trend, seasonality.
Offline advertising
The share of sales generated by offline activity where present, for example local partnerships and events.
Online advertising
The share of sales generated on digital channels: video, TikTok, Meta and the other active channels.
Extra features
The impact of non-media levers: price, promotions, product launches, events, partnerships, competitors.
The model structure
A hierarchical Bayesian structure at country level.
To handle the complexity of three markets the model compares the characteristics of France, Spain and Japan, surfacing where they behave alike and where they follow dynamics of their own.
The same structure makes projections more robust: where data is partial in one market but complete in another, the model still estimates the first with greater confidence, leaning on what it has observed elsewhere. In time, that reading can also inform entry into new markets.
Channels and levers
Awareness, performance, and everything that is not media.
Online campaigns are read in two groups, according to the role they play. Alongside them sit the partnerships, a minimal-investment test on digital-out-of-home circuits and, at the end of the year, a product launch in Spain and France.
Awareness
Programmatic, video, Prime Video and TikTok.
Performance
Meta, retargeting, affiliation and Google, aimed at direct conversion.
Partnerships
A major international sporting event, professional athletes, a broadcaster and the brand’s ambassadors.
CRM as a lever, category by category.
The brand holds a rich dataset on its email communications, already categorised by function: product, brand, promotions tied to a discount, and events the brand runs. The model treats each category as a lever in its own right, to isolate the specific role of each rather than collapsing them into one aggregate reading. That variety of data is one of the project’s strengths: it lets the model capture levers of very different natures.
How we work
A model that is retrained over time.
The model is retrained and recalibrated periodically, so it keeps capturing complex relationships between marketing levers and the results obtained.
Impact of past strategies
Estimates how much each channel contributed to results over the chosen period.
Reading the markets
Captures the differences between France, Spain and Japan and uses them to steer future investment.
Online and offline together
A single reading that also covers activity which cannot be tracked to a click.
What-if scenarios
Simulates new plans before spending, to see the expected effect of a change of course.
Budget allocation
Suggests how to redistribute budget across channels and markets to get more from the same investment.
New markets
The country-level structure can, in time, inform entry into new markets.
Where the project stands
A project in progress.
The project is live and final results are not available yet. Beyond the numbers, the value of the method lies in separating real signal from noise, understanding what actually drives sales rather than what merely accompanies them, and testing budget decisions before committing real spend. It is a method to rely on over time, not a single point figure.