Skip to content
Skip to content

Paid media

How to structure Meta Ads when creative is the targeting

A practical way to plan Meta Ads around offers, events, and a weekly creative test — without treating Advantage+ as a strategy.

By Zinrra ·

Meta will spend whatever you give it. The account structure that used to feel like control — dozens of ad sets, interest stacks, lookalikes stacked on lookalikes — now often fights the delivery system. That does not mean you abdicate. It means the work moves upstream: offer, event quality, and creative tests you can actually read.

What “structure” should mean now

Treat structure as three layers:

  1. Campaign job — prospect, retarget, or retain. One job per campaign.
  2. Conversion event — the action a salesperson would recognise as progress.
  3. Creative hypothesis — the angle you are trying to prove this week.

If a campaign has two jobs, you will not know which one the algorithm optimised for.

Start with the event, not the placement

Before you debate Advantage+ versus manual placements, answer: which event should Meta optimise toward?

  • Purchase, for ecommerce with enough volume.
  • A qualified lead, for services — booked call, application, or a form that includes a disqualifying question.
  • Not “View content” or “every button click”. Those events teach the system to find browsers.

Pixel plus Conversion API should send the same event names. Deduplicate. If sales happen offline, import them. Otherwise you are asking Meta to scale a proxy that your CRM already knows is weak.

Write angles before you open Ads Manager

A useful weekly test is small:

  • One promise (price, speed, proof, risk reversal).
  • Two hooks that say that promise in different ways.
  • One landing section that repeats the winning hook.

Reviews, support tickets, and sales call notes are better source material than the homepage headline. If the site says “innovative platform” and customers say “I finally saw who opened my WhatsApp broadcast”, the ad should say the second thing.

How we run a week

Monday: pick one hypothesis.
Tuesday–Wednesday: ship two or three creatives against it.
Friday: keep, kill, or rewrite. Write the reason down.

Budget still matters. A test without enough spend will look like a coin flip. A test with too many variables will look like insight and be neither.

What not to do

  • Do not launch fifteen ad sets “to gather data” if each one is starved.
  • Do not use a purchased WhatsApp list as a custom audience and call it remarketing.
  • Do not scale a winner by duplicating it into a new campaign every time someone gets nervous. Document the learning; then raise budget in the campaign that already has it.

If you want this system run with you, see Meta Ads or the anonymised paid social rebuild.

Related services

Related work

Keep reading