Paid Media2 min read

A Meta Ads account structure that survives scaling

Most accounts break at the same point: budget goes up, performance goes down, and nobody can tell which change caused it. Here is the structure I use instead.

Abdullah KhalafBy Abdullah Khalaf

Also available in العربية

Scaling a Meta account is less about finding a winning creative than about building a structure where you can tell what happened. Most accounts I audit fail that test - performance drops and there is no way to attribute the drop to a specific decision.

Separate learning from earning

Testing and scaling belong in different campaigns with different budgets and different success criteria. A test campaign exists to produce information at the lowest cost per conclusion. A scaling campaign exists to spend as much as it can while staying above your margin floor. Mixing them means every budget change contaminates your test data.

Three layers, not five

Prospecting

Broad and interest-based, kept deliberately simple. On a store with real purchase volume, broad targeting with strong creative usually beats hand-built interest stacks - the algorithm has more signal than your audience research does.

Catalogue and dynamic

DPA carries more of the revenue than most reports credit it with, because attribution tends to assign the sale to the last touch. Judge it on incremental lift, not on its own reported ROAS.

Retargeting

Small, and shrinking as prospecting improves. A retargeting budget that keeps growing is usually a sign that prospecting is failing to bring in new people.

The rules that keep it readable

  • Change one thing at a time. If you raise budget and swap creative on the same day, you have learned nothing.
  • Give the algorithm room. Budget changes above roughly 20% restart learning. Frequent small increases beat occasional large ones.
  • Consolidate ad sets. Ten ad sets splitting the same budget each get too little data to exit learning. Fewer, larger ad sets converge faster.
  • Name things so a stranger can read them. You will not remember what "Test 3 - v2 final" was in six weeks.

What this looked like in practice

On a Saudi e-commerce account, restructuring from seventeen overlapping ad sets to six - and moving testing into its own campaign - took peak Google ROAS past 500% and let top Meta campaigns hold above 7x while spend increased. Not because the creative got better, but because we could finally see which changes were working.

Related reading

Growth & Analytics2 min read

ROAS is not a strategy

A 78x return told me almost nothing useful. Here are the four numbers that actually decide whether paid media is paying back.

Read article
Automation & AI2 min read

Connecting Odoo to your marketing stack

Marketing dashboards tell you what a campaign spent. The ERP knows what it earned. Joining the two is where automation starts paying for itself.

Read article