Measurement is a design problem

Testing should be rooted in three things: the scientific method, the brand’s own traits, and platform dynamics. Clear hypotheses, measurable goals and rigorous execution are non-negotiable. How elaborate the apparatus should be is not.

A business doing $10M a year through a single sales channel needs a different approach than one doing $100M across several. Incrementality testing might be unnecessarily complex and costly for the first and essential for the second. Platform behavior moves the same way. Changes to auction dynamics and ad delivery quietly rewrite which levers matter, so a measurement plan that was right two years ago may be measuring the wrong thing now.

Measurement is therefore a design problem, and it turns on two decisions.

What you measure

Attribution tells you which channel claimed the sale. Incrementality tells you whether the sale would have happened anyway. Only one of those is a basis for spending money.

The distinction sounds academic until a channel is up for a budget decision. Attribution rewards whichever platform is best at claiming credit, which is not the same as whichever platform created demand, and the gap between them is invisible in a dashboard built on last click.

Building incrementality into the measurement architecture is how a new channel gets validated and scaled to a third of total ad spend on evidence rather than advocacy. The same discipline is what lets you stop paying a channel that has been quietly harvesting demand you already had.

What you measure against

Knowing what you caused is only half of it. The other half is choosing the unit.

Over the last decade consumer brands have leaned harder on paid media to grow while expanding the number of channels they distribute through in order to reach profitability. Most still measure the return on that spend against their online store alone.

That produces a less accurate understanding of the business and handicaps future growth. Spend that looks unprofitable against DTC revenue may be comfortably profitable once wholesale and retail are counted, and a brand that reads the DTC number literally will cut the spend that was building its retail velocity.

The fix is structural rather than analytical:

• Define a combined advertising budget across channels of distribution.
• Consolidate management of direct response, delayed response and retail spend.
• Amortize return across every point of distribution rather than the store alone.

Left separate, each channel optimizes against a number that misrepresents the business, and the org chart quietly becomes the measurement model.

Rigor that earns its cost

None of this argues for maximum apparatus everywhere. The point is to experiment in order to unlock growth, not to test for testing’s sake. Rigor that outruns the decision it informs is just cost. Rigor that falls short of it is how a budget gets set by whoever tells the most confident story.