Stop 6 of 9 · Weekly
Campaign reporting
Read the scheduler and paid exports as a designer, split the two languages before you average anything, and turn the learnings into a next iteration brief with one variable per test.
The report a designer needs is not the report marketing runs
Marketing reports reach, cost and conversions. Farah needs something else out of the same export: which visual decision earned attention and which one cost it. That question is answerable only if the assets were built as a system, which is why the matrix in stop 2 carried the master each asset derived from.
Two files go in. The scheduler export for organic, the paid export for everything with money behind it. Both post-launch, both public numbers, so nothing here needs a code name any more.
Split before you average
Read Arabic and English separately, always, and read format families separately inside each. An in-house campaign that averages well and fails in one language is the most common result in this market and the least visible one in a summary row.
[attach the scheduler export and the paid export] Read these as a creative report, not a media report. Split every figure by language first, then by format family. For each split give me: what performed, what did not, and the size of the gap. Then tell me which differences are large enough to act on and which are inside the noise for the volume we had. Do not recommend anything yet.
One variable at a time
Here is the asset matrix with the master each asset derives from: [paste the matrix]. Group the assets by the single visual variable that differs inside each group: crop, headline length, image treatment, colour block, product on white against product in scene. Compare only within a group. Name the variable that moved the number in each group, and list every group where too much changed at once for the comparison to mean anything.
The second list is the useful one. It tells Farah what to hold constant next time, and it is the reason the following campaign can answer a question this one could not.
| What the export can answer | What it cannot |
|---|---|
| Which format family held attention | Whether the concept was right |
| Which language carried the campaign | Whether the Arabic read naturally |
| Which crop survived the story frame | Whether the brand looks better than last year |
The next iteration brief
From the two reads above, write the next iteration brief for campaign [code name]. Three sections: keep, kill, test. Every keep names the evidence. Every kill names what it frees up in the asset count. Every test names one variable, the two versions, the assets it runs on and the number of results needed before it can be read. Finish with the revised asset count and the formats that change.
Turn that brief into two slides of text: one slide of what we learned, at most four lines, each one a design decision rather than a metric, and one slide of what changes next campaign with the asset count and the date. Write both in Arabic and English. No adjectives about performance, only what we will do differently.