Attribution Isn’t Incrementality
Attribution and incrementality are often used as though they answer the same question. They don’t.
Attribution asks: Which touchpoint gets credit?
Incrementality asks: Would the outcome have happened without the marketing?
That distinction matters because a sale can be attributed to an ad without being truly incremental. A customer may already know the brand, search its name, click a paid ad, and purchase. Search receives the conversion, but the customer may have purchased anyway.
The reverse can also happen. A customer sees a social or video ad, remembers the product, and buys later through a retailer. The media may have created an incremental sale while receiving little or no credit.
This is why platform ROAS isn’t the same thing as marketing ROI. Platform ROAS tells you how much attributed revenue a platform connected to its own spend. It doesn’t automatically tell you:
- What would have happened without the media
- Whether the customer was truly new
- Whether the revenue was duplicated across platforms
- Whether the sale shifted from DTC to retail
- Whether total business revenue increased
- Whether the investment improved profit
One Dashboard Can’t Answer Every Measurement Question
Connecting Meta, Google, DTC, and retail data in one place is a major improvement. But a dashboard is an organizing tool, not a measurement strategy. Different methods answer different questions.
Attribution Best for: understanding observable customer journeys, daily channel optimization, identifying common conversion paths, assigning operational credit. Weakness: it favors touchpoints that are visible and close to the purchase.
Clean rooms Best for: matching media exposure with retailer transactions, measuring retail and in-store activity, working within privacy restrictions, understanding retailer-specific customer behavior. Weakness: they generally operate within a specific retailer or data environment and do not automatically prove causality.
Incrementality testing Best for: determining whether marketing changed the outcome, measuring true lift, testing awareness, CTV, social, or retail media, separating demand creation from demand capture. Weakness: tests require enough scale, careful design, and a willingness to maintain a control group.
Media mix modeling Best for: understanding how channels contribute over time, measuring channels that do not produce direct clicks, evaluating the relationship between spend and business outcomes, guiding larger budget-allocation decisions. Weakness: less useful for daily campaign optimization, and depends heavily on data quality and sufficient history.
The goal isn’t to choose one perfect model. It’s to build a measurement framework that uses the right method for the decision being made.
What Strong Apparel Measurement Should Answer
A useful reporting setup should make it easier to answer questions such as:
- Did marketing create new demand or capture existing demand?
- Which channels introduced customers to the brand?
- Which channels completed the transaction?
- How much revenue occurred through DTC, retail, wholesale, and stores?
- Did retail growth add new revenue or move sales away from DTC?
- Are multiple platforms claiming the same customer?
- Which investments created incremental growth?
- Where should the next dollar go?
- Could the recommendation hold up in a conversation with finance?
That final question matters. Marketing teams aren’t judged on whether every platform report was technically correct. They’re judged on the decisions those reports produce.
Reporting Should Survive the Meeting
The hardest part of reporting usually happens after the dashboard is finished.
Leadership asks whether marketing drove the growth. Finance asks why total revenue didn’t increase as quickly as attributed conversions. Someone asks what happens if the budget is reduced. Then the CMO wants to know which channel should receive the next dollar.
A strong measurement framework shouldn’t eliminate uncertainty by pretending every customer journey can be tracked perfectly. It should, however, help the team make a defensible decision despite the uncertainty.
That means reporting should explain:
- What happened
- Why it likely happened
- What the available data can and cannot prove
- What should change next
- How the recommendation affects revenue and growth
It’s not about making a prettier dashboard, it’s about putting meaning and insight behind numbers. All the data in the world does nothing if it’s not turned into a decision you’re actually willing to defend.
