Meta spends the money. You still steer it.
How we set up a Meta ad account so every test gives a clear answer, every budget follows from your customer cost, and winners can grow without breaking.
Written by Pedram Ghozat, founder of MediaBoostr, and PedramAI · Updated
Why most Meta accounts can’t grow
Most accounts we open are a pile of campaigns nobody can read. The ads might be fine. The setup around them hides which ones are.
Each of these has a simple rule behind it. The rules are below, with the numbers we use.
Structure didn’t die. It changed jobs.
You will hear that account structure no longer matters: put everything in one campaign and let Meta sort it out. We disagree.
Creative matters more than it used to. Meta’s newer delivery systems read the ad itself to decide who sees it, so interest targeting has lost most of its work. That part is true.
But Meta’s system is built to spend your budget. It doesn’t know your margins, your stock or which buyers are new. You decide where the money is allowed to go. Meta decides how to spend it inside those lines. Without a structure you can’t steer budget, read a test or scale a winner.
Give Meta clear lines and it does its part well. Give it none and it spends where spending is easiest.
Six rules from test to scale
The same rules at every spend level. Every budget is a multiple of your cost per new customer (CAC), so the numbers move with your business.
Split only for a business reason
Start with as few campaigns as possible. Every campaign learns from its own purchases, and Meta only moves money inside a campaign. A new campaign needs a reason: a product line with a different margin, a country with its own stock or site, a buyer who never overlaps, a recurring promo.
One concept per test ad set
The test campaign runs on ad set budgets (ABO), the same amount for each, so every new concept gets spend. Each ad set is one persona, one angle and one offer, with 6+ statics or 3+ videos that look different. Existing customers stay out. Names follow Persona | Angle | Offer | launch date.
Fund each test from your CAC
A test ad set needs enough money for about 30 conversions a week at your average CAC. Under that, the result is mostly noise. Plan roughly one new ad per $1,000 of monthly spend, and give most of the test budget to concepts aimed at new customers.
Read the test on the right days
Days 1 to 3, look only at soft numbers: CPM, outbound click rate, cost per add to cart. Let it run past day 7, because people rarely buy on the first view and weekdays sell differently. Judge cold ad sets on cost per new customer and break-even ROAS on total ad spend.
Move winners into a scaling campaign
An ad that has spent at least 20 × CAC above break-even ROAS graduates. It moves, with the same post ID so likes and comments come along, into a separate scaling campaign on campaign budget (CBO). The test campaign keeps running and looks for the next one.
Grow the budget in steps, and the setup with spend
Raise scaling in steps of 10 × CAC a day, only when the ads in it stay profitable after at least 30 × CAC of spend. Keep 2 to 3 new executions of each winner ready before anything gets paused. Add campaigns only when step 1 gives a new reason, never because spend went up.
What goes into each ad set is on creative strategy. How we measure new customers and ROAS across channels is on tracking and attribution.
What the account looks like
A made-up example: a brand that sells fitted T-shirts, spending a few hundred a day. Two campaigns, each ad set one concept, the ads inside it different executions.
Names carry Persona | Angle | Offer | launch date, so you can filter results months later. Example brand and ad sets are made up.
When an ad set wins: leave it alone. New executions of the winning concept go into a new ad set beside it, same name, v2. The full campaign-by-campaign setup is in our Meta ads account structure guide.
Every budget is a multiple of CAC
One number drives the whole setup: what it costs you to get a new customer. These are our rules of thumb, not Meta rules.
Example: CAC $20, break-even ROAS 2.0. At about $400 a day that is two live tests and the start of scaling. Use your own CAC.
Brands spending on Meta who can’t tell what’s working
If the account has grown campaign by campaign and nobody can say which concept brings new customers, structure is where we start.
Questions
Does account structure still matter on Meta?
Yes. Meta’s delivery now leans on the ad to find buyers, so creative matters more. But Meta can’t know your margins, stock or which buyers are new. The structure is how you tell it where money may go, and how you read what worked.
ABO or CBO?
Both, for different jobs. We test on ad set budgets (ABO) so every new concept gets a fair share of spend. Proven ads move to a scaling campaign on campaign budget (CBO), where Meta can push money to the best of them.
How much budget does a test ad set need?
Enough for about 30 conversions a week at your average cost per new customer. With a CAC of $20 that is about $600 a week, or $86 a day, per ad set. These are our rules of thumb.
How long should a Meta test run before I judge it?
At least a full week. In the first three days we only look at soft numbers like click rate and cost per add to cart. Cutting on day two means the sale often shows up later under another ad.
Should I add new ads to a winning ad set?
No. It changes how the ads inside take turns and you lose the clean read. Launch a new ad set next to it with the same name and v2.
How many campaigns should a DTC brand run?
As few as the business allows. Usually one test campaign and one scaling campaign, plus a campaign only where there is a business reason: a product line with a different margin, a country with its own stock, or a recurring promo.