This media mix modeling example shows how a D2C subscription brand read a five percent shift in subscriptions, understood what was driving it, and used that insight to decide exactly where its next dollar of spend should go.
- Sector
- Direct-to-consumer, subscription business
- Scope
- One D2C subscription brand, national-level media data
- Engagement
- Three years, national-level data
Always-ON measurement, delivered on the MassTer platform.
The challenge
First-half subscriptions moved five percent year over year, a shift big enough that the brand needed to understand exactly what was behind it. That shift raised three open questions: how much was media, how much was competitor activity, and how much was something else, like seasonality or a channel running out of room to grow.
Underneath those questions sat a harder one. The brand’s media channels didn’t operate in isolation. Search, social, and partnership activity moved together, and any model that treated them as independent would misstate what each one was actually worth.
The solution
An advanced marketing mix framework was built to separate each driver of the subscription decline, quantify its ROI and cost per acquisition, and give the brand a clear basis for reallocating budget rather than guessing at it. This kind of model separates correlation from cause, the same distinction covered in our incrementality measurement approach.
| Element | Approach |
|---|---|
| Model type | Log-linear (multiplicative), to capture media synergy rather than treat each channel as independent |
| Synergy method | Nested indexed search, to isolate the brand effect within paid search |
| Data scope | National level, three years of media and subscription data |
| Key transformations | AdStock (carry-over), diminishing returns (saturation), weighted sum (simultaneous-channel impact) |
Impressions were used for online channels other than branded paid search, which was measured on clicks; TV on TRPs; out-of-home on insertion counts.
Results and Impact
Online media turned out to explain most of the subscription base, and nested modeling showed Facebook was the single largest individual contributor, ahead of branded and generic paid search. The model also confirmed a real synergy effect: generic paid search performed better specifically when it ran alongside a strong social and partnership presence, not on its own.
45%
of subscriptions traced to online media, the single largest driver of the base
11%
Facebook’s individual contribution, the largest of any single channel
13%
of subscriptions “stolen” by competitor promotional and media activity
−3.9%
subscription loss traced directly to pulling back Outdoor spend
On the other side of the ledger, competitor promotional and media activity explained a real, quantifiable share of the decline. That finding reframed the whole problem: part of the drop was never really lost, it was contested. Positive seasonal effects, mainly summer and school holidays, worked in the brand’s favor. And a pullback in outdoor spend turned out to have a measurable cost.
“The five percent drop looked alarming until the model showed exactly which parts of it were ours to fix, and which parts belonged to the competition.”
A MEMBER OF THE BRAND’S GROWTH TEAM
What changed
- Keep investing in online media, since most channels showed no saturation
- Apply spend control on Instagram specifically, which was starting to show saturation
- Lean into the Facebook and paid search synergy during promotional periods
- Reconsider TV: high spend relative to digital, weak lift, and clear saturation signs, though its brand effect still matters when competitors are actively communicating
- Explore TikTok and YouTube further; neither showed any sign of saturation
Want to see how to capitalize on your own shift?
MASS Analytics builds marketing mix models that separate media, competition, and the calendar so nothing hides. Talk to our team.
Book a demo
