ROAS &amp; Incrementality

The One-Hour Black Friday Claim Ratio Audit

Six steps and one hour produce the two numbers that every Black Friday ROAS figure depends on: the claim ratio and the coverage rate. No vendor, no new tracking, last year's data. The procedure, with an illustrative worked example.

Four CFO Questions About the Black Friday Budget, Answered

The Q4 budget review asks the same four questions every year, and platform ROAS answers none of them. The answers that survive come from coverage, the claim ratio, a qualified holdout and an interval. Scripted, with dates.

One Number for the Black Friday Planning Slide

The Q4 plan usually carries a slide of platform ROAS figures that cannot be compared to each other. Here is the one-line-per-channel replacement, with an illustrative fill, and the reason each field is there.

The Sunk Cost Prison of Last Year's Black Friday Winner

The channel that reported the best Black Friday ROAS last year is about to get the biggest budget this year, unexamined. Two experiments explain why that is the most expensive habit in the plan.

Black Friday Discount Depth: The Loss You Book in Advance

A 30% discount does not cost you 30%. It moves your break-even ROAS from 2.0 to 3.5 on a 50% margin, before a single ad runs. The arithmetic, in a table you can rebuild with your own numbers.

Black Friday ROAS: What Platforms Report vs Orders Shipped

Meta says one ROAS, Google another, the store a third. The Price of Being Found lists four properties a defensible number has, and platform-reported Black Friday ROAS has none of them.

The Last Day to Start a Holdout Test Before Black Friday

A holdout that cannot finish before Cyber Week cannot inform the Cyber Week budget. Count back from 27 November with the standard design and the last start date is Friday 2 October.

Your Black Friday Ad Budget Is Finite. The Claims Are Not

Every platform will claim the same Black Friday order. Your budget cannot be spent twice. The claim ratio is the one number that separates the two before you allocate.

Sep 7, 2026

Your Landed Cost Went Up. Your ROAS Target Didn't.

Tariffs, freight and supplier switches move contribution margin, which moves break-even ROAS. Most brands never re-run the arithmetic - and the ROAS they compare against is inflated to begin with.

Cut a Channel the Right Way: Run the Holdout First

The safest way to cut a channel is to prove, first, that revenue survives without it.

Rising Ad Costs Are Not the Threat. Misallocation Is.

Everyone blames rising CPMs. The bigger leak is paying those CPMs on channels that never earned the sale.

Chasing a Cheaper Click Is the Wrong Fight

A cheaper click is only progress if the click was ever going to cause a sale. Often it was not.

Are Post-Purchase Upsells Incremental Revenue, or Just Reshuffled?

A one-click upsell raises average order value. Whether it raised your actual profit is a different question.

Free-Plus-Shipping Offers: Measuring If the Front End Actually Pays

A free-plus-shipping offer buys you a customer cheaply. Whether it buys you a profitable one is the real test.

CTV and YouTube Measurement Without a €50K Brand-Lift Study

CTV spend went self-serve, but measurement advice stayed enterprise. Here are the realistic options ranked by minimum viable budget, and which ones are evidence versus theater.

POAS vs ROAS: Attribution When Tariffs Eat Your Margin

Tariff-driven COGS inflation means two campaigns with identical ROAS can produce opposite profit outcomes. POAS puts contribution margin inside the measurement, and the budget ranking flips.

Is Your Branded Search Actually Incremental? The Retargeting Cannibalization Test

Branded search and retargeting routinely claim the same order, and both platforms book it as performance. Here is a 30-minute GA4 export test that quantifies the overlap, plus the honest limits of pattern detection.

The Coupon and Affiliate Incrementality Audit: Stop Paying for Sales You Already Had

Coupon extensions and last-second affiliate intercepts claim credit at the point of sale, and networks book it as performance. This audit quantifies the non-incremental share from a plain order export, no holdout needed.