One Number for the Black Friday Planning Slide

The Black Friday planning slide needs one line per channel and five fields on that line: incremental ROAS, its interval, the design that produced it, the smallest effect that design could have seen, and the coverage rate of the data underneath. That format is The Price of Being Found's four properties of a defensible number turned into a table, and a plan written in it does not get argued with, because every argument the room could raise is already answered in a column.

Octalysis calls the one obvious next action a Glowing Choice: when everything else on the screen is dim, the thing to do is the thing that glows. On a slide of platform ROAS figures nothing glows, because none of them can be compared to another. This format makes one column glow, and it is the interval.

The template

Every figure in that fill is illustrative and marked as such. The structure is the point.

Why each field is there

Incremental ROAS, not reported ROAS. Reported ROAS comes from the party selling the media and fails the first property before the second is reached. The Black Friday ROAS your platforms will report is the long version.

The interval. The book's example of the wrong sentence is "the channel delivered a 12% lift". The right one is "12%, with an interval of 4 to 20, from a design whose minimum detectable effect was 8.3%". The interval is what stops a point estimate being read as a fact.

The design. Experimental, quasi-experimental, or observational, and if observational, a description rather than a causal claim. Most arguments in marketing meetings, the book observes, are two people comparing an experimental number to an observational one without either noticing. This column ends those.

The MDE. The smallest lift the design could have detected. Without it, an interval that includes zero reads as "does nothing" when it may mean "could not have seen it". Can a brand your size measure Black Friday lift at all? is where the number comes from.

Coverage. Revenue of X, of which the systems can attribute Y. Every estimate on the slide sits on the visible share, and the slide says how big that share is instead of hiding it.

Anchor date. The date of the last qualified holdout. Everything downstream of it is model, not measurement, and confidence decays from it. A row that says "never" is honest and, for most organisations, the book says, accurate.

The row that says "not measurable"

The most valuable line on the slide is the one that declines to give a number. A channel whose spend share times honest return sits below the floor cannot be measured at current scale, and the slide says so and manages it on stated judgement. That row is what converts the plan from a set of confident figures wrong by an unknown factor into an account of what is known and what is not. The book's Monday protocol splits every channel into two lists for this reason, and calls the second list the more useful one.

What to do this week

• If you have to defend the number: rebuild last year's Q4 slide in this format. Most rows will be observational or "never". That is the true first result and the reason to run one qualified holdout before 2 October.
• If you own the budget: ask for the plan in this format and refuse the other one. It takes one sentence and it changes what gets measured.

The calendar has the dates. A causal read on the GA4 export fills the observational rows with intervals, and the demo at demo.causalityengine.ai shows the output on a sample store with no signup.

As of 9 September 2026. The four properties and the anchor-date discipline are from The Price of Being Found (Edition 2.10), Chapters 19 and 20. The table fill is illustrative.