Open a Meta ad account that has been managed by several people over a few years and you will often find the same thing: 20 or 30 active ad sets, each with a slightly different interest stack, age band or lookalike percentage, each spending $15 a day and each stuck in "Learning limited". Every ad set was created for a sensible reason. Together, they make it almost impossible for Meta's delivery system to learn anything.
Meta's own direction of travel is clear. Targeting has moved toward broad audiences and Advantage+ automation, budgets are increasingly set at campaign level, and creative has become the main lever advertisers control. Semrush's 2026 PPC strategy guide sums up the shift: Meta's Advantage+ system is designed to automate audience targeting, budget allocation and optimisation, and such tools need conversion data to work.
That does not mean "one campaign, one ad set" is right for everyone. This guide explains why consolidation usually helps, how to calculate whether your ad sets have enough data, when to use campaign or ad set budgets, the situations where splitting still makes sense, how to structure creative testing, and a step-by-step way to restructure an existing account without losing a month of performance.
Why fragmented accounts starve Meta's delivery system
Meta's auction decides, for every impression, which ad to show to which person and at what bid. Its predictions depend on feedback: the conversions your ads generate. Each ad set is, in effect, a separate learning unit. When budget and conversions are spread thinly across many ad sets, each one learns slowly, stays volatile and costs more per result.
Fragmentation also creates self-competition. When several ad sets target overlapping people, they can enter the same auctions. Semrush's explainer on audience overlap describes the remedy many advertisers reach for:
“The other option is to merge or turn off overlapping ad sets.”
— Semrush, Facebook Audience Overlap Explained + Ways to Avoid It
Our companion guide on Meta audience overlap covers how to measure that overlap. The structural point here is simpler: every split you create needs a reason strong enough to justify dividing your data.
Learning phase arithmetic: does each ad set have enough data?
Meta puts new or significantly edited ad sets into a learning phase while the system explores who responds. Semrush notes that the learning phase "concludes after the ad is shown to enough people for the performance to stabilize". Meta's long-standing guidance is that an ad set generally needs about 50 optimisation events within a week of its last significant edit to exit learning; if it cannot get there, Ads Manager shows "Learning limited". Google uses a similar benchmark for its own social-style campaigns, which is a useful sanity check:
“up to around 50 conversion events or 3 conversion cycles”
— Google Ads Help, About Demand Gen performance during the learning phase
You can test any structure against that threshold with one line of arithmetic: weekly conversions per ad set = (weekly budget ÷ number of ad sets) ÷ cost per result.
A worked example
Hypothetical example. An online store spends $6,000 a week on Meta at a $40 cost per purchase.
| Structure | Ad sets | Budget per ad set per week | Purchases per ad set per week | Likely status |
|---|---|---|---|---|
| Fragmented | 12 | $500 | 12.5 | Learning limited |
| Moderate | 4 | $1,500 | 37.5 | Borderline |
| Consolidated | 2 | $3,000 | 75 | Exits learning |
| Consolidated, optimising for add to cart | 12 | $500 | About 50 (at $10 per add to cart) | Exits learning, but optimises for a weaker signal |
The last row shows the alternative many advertisers take: optimise for a higher-funnel event to get more volume. That can work for small budgets, but the system then finds people who add to cart, not people who buy. Consolidation is usually the better fix when budget allows.
Advantage campaign budget versus ad set budgets
Meta lets you set the budget at campaign level (Advantage campaign budget, formerly called campaign budget optimisation) or at ad set level. Campaign budgets let the system move money toward the ad sets that are getting cheaper results right now; ad set budgets give you fixed allocations.
| Question | Advantage campaign budget | Ad set budgets |
|---|---|---|
| Who decides the split between ad sets? | Meta, in real time | You |
| Best when ad sets… | Share a goal and a similar value per conversion | Have different goals, values or must-spend requirements |
| Risk | One ad set can take most of the budget; small ones may barely spend | Money stays in weak ad sets until you move it |
| Control options | Ad set minimum and maximum spend limits | Manual changes and rules |
| Typical use | Prospecting with several creative-led ad sets | Tests that need equal spend; retention budgets; country budgets |
A common mistake is to use a campaign budget across ad sets with very different purposes, such as prospecting and retargeting, and then be surprised when retargeting takes most of the money because its cost per result is cheaper. Cheap is not the same as incremental. Keep ad sets inside a campaign-budget campaign comparable, or put caps on the ones that would otherwise dominate.
Bid strategies interact with structure
Meta's default is to spend the budget for the most results it can get. Cost-per-result goals, ROAS goals and bid caps add a constraint, and constrained ad sets often under-spend. That changes the consolidation maths: if an ad set with a cost-per-result goal spends only half its budget, its effective data volume halves too. Our rule of thumb is to consolidate first on the default strategy, and to add cost or ROAS goals only once an ad set is producing stable volume, setting the goal close to the cost you are already achieving rather than an aspirational target. If you do use goals, keep them consistent across the ad sets inside a campaign budget so the system is comparing like with like.
The same logic applies on Google: its Performance Max channel report notes that "you can't directly control budget allocation per channel", only influence it. Automated allocation is efficient when the units being compared are genuinely comparable.
A consolidated account blueprint for 2026
For most ecommerce and lead-generation advertisers, a structure along these lines is a sensible default. Treat it as a starting point and change it for the reasons in the next section.
- Campaign 1: Prospecting (sales or leads objective). Broad or Advantage+ audience, with existing customers excluded or capped where the platform allows. One to three ad sets at most, each built around a creative theme rather than an audience slice. Campaign budget.
- Campaign 2: Retention or remarketing, only if you have the volume and a different message (for example a returning-customer offer). Kept small, with clear exclusions.
- Campaign 3: Creative testing. A controlled place to introduce new concepts before they graduate to Campaign 1 (see below).
Meta's Advantage+ sales and leads campaigns can replace Campaign 1 for many accounts. Our guide to Advantage+ campaigns covers when they work. Targeting options in a broad setup are covered in our guide to Meta ads targeting.
When splitting campaigns or ad sets still earns its place
Consolidation is a default, not a rule. These are the situations where a separate campaign or ad set is justified:
- Different conversion events. A lead campaign and a purchase campaign optimise toward different signals and should not share a budget.
- Different countries with different economics. If a customer in one country is worth twice as much, or the landing pages and language differ, split by market.
- Different value per conversion. A $30 accessory and a $900 product line can share a campaign only if you optimise for value, not volume.
- Must-spend commitments. A product launch, a retailer co-op budget or a regional franchise that has paid for its own spend needs a guaranteed allocation.
- Special ad categories. Housing, employment, credit and social issue ads have targeting restrictions and must sit in correctly declared campaigns.
- Clean experiments. A/B tests need controlled splits; Meta's test tools create them for you.
- Genuinely different messages to known groups. Existing customers seeing a loyalty offer versus new prospects seeing an introductory offer.
| Scenario | Split? | Better alternative if not |
|---|---|---|
| UK and US markets with different prices and pages | Yes, by campaign | — |
| Men versus women for a unisex product | No | Gender breakdown in reporting; creative variety |
| Three interest stacks "to see which wins" | No | One broad ad set; test creative concepts instead |
| Lead form ads and website purchase ads | Yes | — |
| 1% and 5% lookalikes of the same seed | Rarely | One ad set with the wider audience or Advantage+ audience |
| Feed and Stories placements | No | Placement-specific assets within one ad; placement breakdown |
| Ad set for each new creator video | Only in a testing campaign | Graduate winners into the main ad set |
What does not usually justify a split: slicing by age band, gender or a single interest "to see what works". Meta's reporting breakdowns show performance by age, gender, placement and region without dividing your data. Our guide to reading Meta Ads Manager reports explains how.
Structuring creative testing so winners can scale
With audiences largely automated, creative is where structure decisions now matter most. There are two workable models.
Model A: test inside the main campaign
Add new ads to an existing prospecting ad set and let delivery decide. It is simple and keeps data together, but Meta tends to favour ads with a performance history, so new ads may get little spend and you learn little.
Model B: a dedicated testing campaign
Run a separate campaign with a fixed, modest budget where each new concept gets its own ad set (ad set budgets for equal spend) or uses Meta's A/B testing. Concepts that beat your benchmark cost per result after a pre-agreed spend are copied into the main campaign, keeping the post ID where possible so engagement carries over.
Most accounts spending more than a few thousand a month benefit from Model B, with 10 to 20% of budget assigned to testing. What you test matters more than how: test concepts and hooks before small variations. Google's ABCDs framework for video, built on its research with Kantar, is a useful checklist for what a concept should include. Our full framework for Meta creative testing goes into decision rules, and formats such as carousel ads, UGC ads and partnership ads are natural candidates for the testing campaign.
Restructuring an existing account without crashing performance
Turning off 25 ad sets on a Monday and launching three new ones usually leads to a week or two of unstable results while the new ad sets learn. A staged migration reduces the risk:
- Audit the last 60–90 days. List each ad set with spend, results, cost per result and learning status. Identify the top ads by spend and results; these are your proven creative.
- Fix measurement first. Confirm that the pixel and Conversions API are sending the events you optimise for, with good match quality. Restructuring on broken data wastes the effort. See our pixel and Conversions API guide.
- Build the new structure alongside the old one. Launch the consolidated campaign with the proven ads (using existing post IDs to keep social proof) at around 20–30% of total budget.
- Shift budget in steps. Each week, move a further share of budget from the old ad sets to the new campaign, pausing the weakest old ad sets first.
- Avoid frequent edits to the new campaign. Large budget swings and constant changes restart learning. Make budget increases in modest steps.
- Judge on blended results. Compare total account cost per result and revenue week on week, not the new campaign alone.
- Pause the old structure once the new campaign is out of learning and blended results are stable or better. Keep the old campaigns paused, not deleted, for reference.
Expect some noise. Semrush's Facebook ads troubleshooting guide treats a week of rising cost per result as a warning sign worth investigating, not an emergency; the same patience applies during a restructure.
Questions advertisers ask about Meta campaign structure
How many ad sets should a Meta campaign have?
As few as your goals allow. A useful check is whether each ad set can reach roughly 50 optimisation events a week; if not, consolidate or use a higher-volume event.
Is Advantage campaign budget always better?
It works best when ad sets share a goal and similar value. Use ad set budgets for tests that need equal spend or for must-spend allocations.
Should retargeting have its own campaign?
Only if you have enough volume and a genuinely different message. Broad prospecting campaigns already reach many past visitors.
Does consolidation hurt reporting?
No. Use breakdowns by age, gender, placement and region instead of separate ad sets for each segment.
How long should I wait after restructuring?
Give new ad sets at least a week or two, and longer if conversions are slow, before making major changes.
Simplifying your Meta account structure this month
Run the learning-phase arithmetic on your current account. If most ad sets cannot reach the threshold, plan a staged consolidation: fix tracking, build a simpler prospecting campaign with your proven ads, move budget in steps and set up a separate testing campaign for new creative. Keep splits only where goals, markets, values or legal categories genuinely differ.
If you would like specialists to audit and rebuild the account, our social media management team runs Meta campaigns for ecommerce and lead-gen clients, and our PPC management service covers cross-channel budgets. Get a free quote or request a custom quote.
Further reading on meta ads campaign structure
- UGC Ads: How to Brief Creators for Ads That Convert
- Google Ads Disapprovals and Account Suspensions: How to Fix Them
- Disabled or Restricted Meta Ad Account: How to Recover It
Sources for this Meta campaign structure guide
- How to build a PPC strategy that gets results (Semrush, July 2026)
- Facebook Audience Overlap Explained + Ways to Avoid It (Semrush, October 2023)
- 8 Signs of Facebook Ads Issues and How To Fix Them (Semrush, 2020)
- About Demand Gen performance during the learning phase (Google Ads Help)
- About the channel performance report for Performance Max (Google Ads Help)
- About the ABCDs of effective video ads (Google Ads Help)
