Stop 5 of 9 · Weekly
Meta: creative fatigue and the test plan
Read fatigue out of the Meta export, size a test plan that can actually resolve, and brief the designer in one page.
What it is
Meta creative dies on a schedule and Nour clients notice the week after she does. This stop turns the Ads Manager export into three things: which ads are past their useful life, what the next test should be, and a brief the designer can build from without a meeting.
Reading fatigue out of the export
Pull the export broken down by ad and by week, with impressions, reach, frequency, CTR, CPM, results and cost per result. One breakdown is worth adding: placement, because an ad that looks tired in aggregate is sometimes just a Reels problem.
The signal is a sequence, not a threshold. CTR falls week on week while frequency climbs, then cost per result rises roughly a week later. Frequency above some number is an argument with no end. The sequence tells you which week the ad turned, and that week is what you multiply by daily spend to get the money.
[upload the Meta export by ad and week with frequency, CTR, CPM, results and cost per result] For each ad, find the first week where CTR fell while frequency rose, and treat that as the turn week. Show cost per result before and after the turn, and the spend since the turn multiplied by the difference, as wasted spend. Rank the ads by wasted spend. Recommend three to pause, two to refresh with the same offer and new visuals, and any winner worth moving into a new audience. Show the weekly rows behind the top three. Change nothing in the account.
Sizing the test
A test plan that cannot resolve is a slower way of guessing. Before the designer starts, Nour fixes the number of concepts, the one variable each changes, the budget per cell and the results each cell needs before anyone reads it. On a local lead budget that usually means two or three concepts, not six.
From the fatigue analysis, write the test plan for the next four weeks. Three concepts maximum. For each: the hypothesis in one sentence, the single variable it changes against the current control, the daily budget it needs, and the number of results it must reach before I read the outcome. Then state the earliest date I can call it, and what result would make me stop it early.
The brief to the designer
One page, no adjectives that cannot be built.
Turn concept two of that plan into a brief for our designer. Include: what we are testing and against what, the audience in one line, the offer and the proof point, the exact primary text, headline and description in three lengths, the visual direction described as things to show rather than moods, the aspect ratios and placements it must work in, and what must not appear for compliance in this vertical. Keep it under one page. Mark anything I need to confirm with the client before it is built.
The measurable
Two numbers per account per week: wasted spend since the turn week, and the number of live tests that are correctly sized. The first should fall. The second should almost never be more than three.