Meta Ads Campaign Structure: How Many Campaigns Do You Need

Most accounts are not badly optimised. They are badly divided. The structure question comes before every setting inside it.

The default that works

One campaign per objective, per market.

CONV_US    sales, United States
CONV_UK    sales, United Kingdom
LEAD_US    leads, United States

A campaign exists to separate what must not share a budget or an optimisation goal. Sales and leads must not. Two markets with different costs per result usually must not. Two creative angles absolutely must not: that is what ad sets are for.

Where the budget lives

Under CBO, the budget sits at campaign level and Meta allocates between ad sets. You get the best delivery and the least control.

Under ABO, each ad set has its own budget. You get a guaranteed floor per ad set and worse global allocation.

The choice is not ideological. During a structured test, ABO protects the ad sets that would otherwise be starved before producing a read. In steady state, CBO usually wins.

The mistake that costs the most

Splitting one audience across several campaigns. The auction does not know these campaigns belong to you: they bid against each other, and the increased competition shows up in your CPM.

If two campaigns target the same people, they should be one campaign with two ad sets.

How many ad sets

As many as the budget can feed. Each ad set needs roughly fifty optimisation events in a rolling seven day window to exit the learning phase.

Ten ad sets on a budget that produces sixty conversions a week means no ad set ever exits learning. The structure looks rigorous and produces nothing you can trust.

The ad level, which most structure advice skips

Campaign and ad set get all the attention. The ad level decides something just as structural: how many ads sit inside one ad set, and what each one contains.

Meta concentrates delivery on the two or three ads it likes best inside an ad set. Put eight ads in one ad set and you have not run an eight way test, you have run a two way test and paid for six ads that barely served.

The second decision is what counts as one ad. A concept usually exists in four ratios, and those four files belong inside a single ad so the right ratio serves per placement. Splitting them into four ads multiplies your ad count, splits the delivery signal four ways, and makes the reporting harder to read for no gain.

Ads per ad set	What you actually learn
1	A clean read on that creative, if the budget supports it
2 to 3	A usable comparison, delivery still concentrates
6 or more	Which two Meta preferred, and nothing about the rest

Testing campaigns, kept separate

CONV_US_CBO_Evergreen
CONV_US_ABO_Testing

Two campaigns, one purpose each.

The evergreen campaign holds what works and is left alone. The testing campaign holds what is being tried, under ABO so that each test gets a guaranteed floor rather than competing for scraps of a shared budget.

Keeping them apart has a second benefit that shows up at reporting time: exploratory spend does not contaminate the numbers you present as account performance.

When a new campaign is genuinely justified

Create one when something at campaign level must differ:

• A different objective, since sales and leads cannot share optimisation
• A different market with materially different costs
• A budget that must not be shared, typically a promotion with its own envelope
• A different budget mode, CBO against ABO

Create one for none of these reasons and you have added a competitor to your own auction.

The audit worth running once a quarter

Open the campaign list and ask three questions.

Which campaigns target the same audience. Which have not spent in thirty days. Which exist because of a promotion that ended.

Most accounts carry between three and eight campaigns that should have been paused months ago. They are not costing much directly, and they are making every list, every filter and every export harder to read.

What this does not solve

Structure decides what your numbers mean. It does not decide whether the creative is any good, and no arrangement of campaigns rescues an offer nobody wants.

It also does not survive contact with a live account for long. Structures drift as people add campaigns for one-off promotions and never clean them up, which is a governance problem rather than a media buying one.