Finding weak ad spend before the budget freeze

Once the budget is locked, a finding is a note for the following cycle. The weeks before the freeze are the only window where a read changes an allocation rather than documenting one.

What fits before the freeze

The six-week version

A causal read to narrow the field, then a geo holdout on the single channel where the read and the platform reporting disagree most. That produces one properly evidenced number in time for the plan, which is worth more than a table of unresolved estimates.

Holdouts need a pre-period, so six weeks is close to the minimum. The last day to start a holdout covers the same arithmetic for the peak calendar.

The two-to-four week version

A read on a window with real spend variation in it, plus the claim ratio. That is enough to argue for a different weighting and to explain why the plan should not be built directly on platform-reported returns.

The read is €99 once on a Google Analytics export, refundable within 30 days, no questions asked, returning per channel what it caused next to what last-click gave it, a data-health score rather than a confidence interval, and a next step.

The under-two-week version

Just the claim ratio. Twenty minutes, no tooling, and it reframes how the room reads the platform numbers it is about to allocate against. It will not change the split this year and it often changes how next year's process is designed, which is the larger win. Method in the claim ratio.

The one thing to secure in the plan itself

Permission to hold one channel steady for a measurement window during the year. It costs nothing at planning time and it is nearly impossible to get later, because by then the money is committed and any pause looks like a cut.

That single line in the plan is what makes next year's read resolvable, and almost nobody asks for it.

Check before you plan a window

Your analytics retention setting governs how far back you can read, and it expires quietly. Confirm it covers the period you intend to analyse before building the plan around it, as covered in the two-month retention trap.

The honest note

The freeze is a real deadline and it is the only one here. Everything else can be done afterwards, and doing the read afterwards is still worth it, because next year's plan starts sooner than it feels.

Related answers

• What looks like wasted ad spend but is not
• What instant channel diagnosis can and cannot do
• Confirm a weak channel with a holdout test
• Automate attribution before the Black Friday window
• The reported vs causal ROAS gap widens at peak