Structure is a data decision
It is tempting to treat account structure as housekeeping, a matter of tidy names and logical folders. On Meta it is something more important. Every split you make divides the conversion data available to the delivery system. Divide it too finely and no part of the account has enough signal to learn, so results become erratic and expensive.
The framework below starts from that constraint. Structure should be as simple as the business allows, and every split should be justified by a real difference in goal, economics or audience, not by a wish to see more rows in a report.
The three levels and what each controls
Fig. 01 · Stack
Tap to explore
Meta's account hierarchy
Campaign
Objective. Optionally budget, if you let Meta distribute it across ad sets.
Ad set
Audience, placements, schedule, optimisation event, budget if set here.
Ad
Creative, copy, format, destination and tracking parameters.
Meta allows budgets to be set at campaign level, where the system moves money between ad sets based on performance, or at ad set level, where you fix the split yourself. Naming conventions for these settings have changed over time, so check current documentation. Campaign-level budgets suit ad sets that are alternatives to one another. Ad set budgets suit cases where you must guarantee spend on a specific audience or region.
When to split, and when not to
A useful rule: split when the business would make a different decision about the two halves. If prospects and existing customers are worth different amounts, or one country has a different margin from another, separate them. If two interest groups would get the same ads, the same budget logic and the same target cost, they probably belong in one ad set.
| Reason to split | Usually justified? | Why |
|---|---|---|
| Different objective (e.g. leads vs sales) | Yes | Different optimisation events need separate campaigns |
| Different country or currency economics | Usually | Targets and budgets differ |
| Prospecting vs retargeting | Often | Different message, frequency and value |
| Different product lines with different margins | Often | Different affordable cost per result |
| Different interest groups, same offer | Rarely | Splits data without changing decisions |
| Different age bands or placements | Rarely | Delivery already optimises across them |
| One ad per ad set for testing | Rarely | Fragments learning; test within ad sets instead |
A consolidation-first framework
Most accounts benefit from a simple base structure, with additional campaigns only where a clear reason exists. The framework below suits a typical D2C or lead-generation advertiser; adapt the names to your business.
Compare scenarios
The three-campaign base
The main engine. Broad or lightly guided audiences, optimised for the core conversion, holding most of the budget.
- One or a few ad sets with ample budget
- Many creative angles inside each ad set
- Existing customers excluded where the goal is new buyers
A smaller campaign where new creative concepts are tried before graduating to prospecting.
- Fixed budget so tests get fair spend
- Clear hypothesis for each concept
- Winners move to prospecting; losers are retired
Reaches recent visitors and engagers with a message suited to people who already know you.
- Small relative to prospecting
- Watch frequency closely
- Judge on incrementality, not only reported ROAS
Some advertisers fold retargeting into prospecting and let delivery decide, especially as Meta's automated campaign options increasingly blend audiences. That can work, provided you can still see whether new customers are being acquired. Our retargeting guide covers the trade-offs.
Matching structure to scale
Fig. 02 · Matrix
Tap to explore
Choosing how much to split
The less data you have, the more you should consolidate. A small account split into ten ad sets is ten experiments that never finish. The same budget in one or two ad sets gives the system a chance to find what works.
Where creative testing fits
In a consolidated structure, ads compete within ad sets, and Meta shifts delivery toward those that perform. That is efficient but can be unfair to new ideas, which may get little spend before the system settles on a favourite. A dedicated testing campaign with fixed budgets gives each new concept a fair hearing.
Test concepts, not trivia. A concept is a distinct reason to buy, aimed at a distinct buyer situation: 'saves time for busy parents' versus 'healthier than the alternative'. Changing a button colour is not a concept. Meta's own experiment tools can help with controlled tests; check what is currently available.
Automated campaign options
Meta increasingly offers campaign setups in which audience, placement and creative combinations are largely automated, especially for sales and app objectives. Names and capabilities change, so check current Ads Manager documentation. These options are the logical end of consolidation: one campaign, broad delivery, many creatives, with the system making most allocation decisions.
They can work very well for advertisers with strong conversion signals and plenty of creative. They are less suitable when you need to guarantee spend on a particular audience, separate new-customer acquisition from repeat purchase, or run controlled tests. A common approach is to run an automated campaign alongside a smaller manual structure, compare them on total business results, and shift budget toward whichever earns more incremental revenue.
Whatever you choose, keep the ability to answer one question: how many genuinely new customers did this spend produce? If the structure hides that answer, it is the wrong structure, however efficient the reported numbers look.
Naming, so humans can read it
A naming convention costs nothing and saves hours. Use a fixed order of attributes separated by a consistent character, for example market, objective, audience type, concept and date. Names should let anyone filter reports by any attribute without opening settings. Pair this with consistent UTM parameters on destination links so analytics tells the same story as Ads Manager.
Checking your current structure
Self-diagnostic
0/6Structure diagnostic
Answer for your main Meta account.
01Do most of your ad sets exit the learning phase?
If yes: Your structure gives the system enough data. If no: Consolidate ad sets until each gets enough optimisation events.02Can you explain the business reason for every campaign?
If yes: Good. The structure reflects decisions. If no: Merge campaigns that exist only for neatness or habit.03Do ad sets in the same campaign target overlapping people with the same ads?
If yes: They compete with each other; merge them. If no: Good.04Does each prospecting ad set hold several genuinely different creative angles?
If yes: Good. Delivery has options to find different buyers. If no: Add concepts before adding ad sets.05Are existing customers excluded from new-customer prospecting where appropriate?
If yes: Your acquisition numbers are more honest. If no: Add exclusions, or measure new versus returning customers separately.06Could a new team member understand the account from names alone?
If yes: Good. If no: Adopt a naming convention and rename as you go.
Structure follows strategy
No structure rescues weak creative, poor tracking or an offer the market does not want. Get those right with the help of our Meta ads guide, then use structure to give the system the cleanest possible data. As platforms add automation, structure becomes less about controlling delivery and more about feeding it clear goals and diverse creative. Keep an eye on creative fatigue, because consolidated ad sets rely on fresh ads to keep reaching new people.
Key takeaways
- 01Every split in a Meta account divides the conversion data the system learns from.
- 02Split only when the business would make a different decision about each part.
- 03A simple base of prospecting, testing and retargeting suits most advertisers.
- 04The less data you have, the more you should consolidate.
- 05Test distinct creative concepts with fair budgets, and name everything consistently.
Frequently asked
- How many ad sets should a Meta campaign have?
- As few as your strategy allows. Each ad set needs enough optimisation events to exit learning, so the right number depends on budget and cost per result. Many accounts perform best with one to three ad sets per campaign, each carrying several distinct creative concepts.
- Should I use campaign budget or ad set budget on Meta?
- Use campaign-level budgets when ad sets are interchangeable and you want Meta to allocate spend to the best performers. Use ad set budgets when you must guarantee spend for a particular audience, region or test. Check current Ads Manager documentation, because naming and options change.
- How many ads should be in each ad set?
- Enough to give delivery real choices, but not so many that each gets too little spend to judge. Several distinct concepts per ad set is a sensible starting point. Check Meta's current guidance for any recommended limits, and retire clear losers regularly.
- Should prospecting and retargeting be separate campaigns?
- Often, because they carry different messages, frequencies and values. Some advertisers combine them and let Meta's automation decide. If you combine them, measure new-customer acquisition separately so retargeting does not flatter results.
- Why do my Meta ad sets never exit learning?
- Usually because each ad set gets too few optimisation events. Common causes are too many ad sets sharing a small budget, optimising for a rare event, or frequent edits that restart learning. Consolidate, optimise for a more frequent event temporarily, or raise budget.
Published by Fabulous.Media, a network of specialist marketing agencies. Updated 9 October 2026. Platform features change often; check current official documentation before acting on platform-specific detail.
