The reported vs causal ROAS gap widens at peak

Reported ROAS looks best during peak and means least during peak, and the two facts have the same cause. The window that most needs a reliable number is the window where the reported one is least reliable.

Three structural reasons

The third is the one that catches brands out. A shopper who would have bought in December buying in November is not incremental revenue, it is the same revenue earlier, at a discount. Reported ROAS cannot distinguish the two, because it sees a conversion attached to a touch.

Why the overlap grows

During peak you increase frequency across channels, so the same shopper sees more of your marketing before buying. Every platform that touched the path claims the order under its own rules, so the sum of claims grows faster than actual revenue. The claim ratio you measured in an ordinary month is not the claim ratio during the window.

That is worth measuring specifically rather than assuming. Run the ratio on a normal month and again on the window, and the difference is the peak-specific overlap. The method is in platform-reported ROAS against orders.

What to hold on to during the window

Your baseline. A peak read without a normal-month comparison is close to uninterpretable, because peak is strong for reasons unrelated to your channel mix. If you have not taken a baseline read, that is the highest-value hour available before the window opens.

Also hold your definitions. Changing channel grouping or lookback mid-window makes before and after incomparable, and that damage cannot be repaired afterwards. The calendar for all of this is in Black Friday measurement deadlines.

The read that is worth most

The post-window read, taken in the two weeks after peak. Budget moved enough during the window to separate channels more clearly than in an ordinary month, so estimates are usually tighter. The constraint is your analytics retention setting, which can expire and take the granular data with it, as covered in the retention trap.

Causality Engine produces that read from a Google Analytics export at €99 once, with per-channel confidence intervals, coverage and a design label, refundable if it does not move a budget decision. The interactive demo shows the output shape with no signup.

What not to do in the window

Do not reallocate large amounts of budget mid-window on reported ROAS, because that is the number most distorted by the conditions above. Small tactical moves are fine. Structural decisions belong to the post-window read, when you can compare against a baseline.

The honest framing of the urgency

Nothing about this is urgent because of a countdown. It is urgent because retention expires, baselines cannot be taken retroactively, and definitions changed mid-season cannot be un-changed. Those are calendar facts and they are the only real deadlines here.

Related answers

• Where reported and causal ROAS agree more than you expect
• Why platform ROAS and causal ROAS disagree
• How to test the gap between reported and causal ROAS
• Attribution updates in Slack during peak week
• What a founder can set up before peak, alone