Testing, Winners and Graveyard: the three-campaign Meta ads structure we run in 2026

I get added to more than 20 ad accounts a week. When one can't scale, the cause is nearly always on a short list: interest targeting, too many campaigns, awareness campaigns on small budgets, or not enough creative.

A tidy structure fixes the first three. The fourth is a creative problem, and no campaign layout solves it. One client came to us spending $1,000 a day and couldn't get Meta to spend a cent more. We changed nothing in the ad account and gave it better ads. Within two weeks it was spending $5,000 a day profitably.

So read what follows as plumbing. It's how we structure Meta accounts for our media buying clients in 2026, adapted from ScaleSurfer's 2026 campaign structure diagram. It replaces the CBO setup I described in my 2025 videos, and the last section explains the change.

The structure on one page

Three campaign types. Ads only ever move one way through them.

TESTING
  ABO, one campaign per product
  one ad set per creative batch,
  3 to 5 ads
    weak signals ........ switched off
    good, but can't scale .. GRAVEYARD
    winner ................... WINNERS

WINNERS
  bid cap, inflated budget,
  one campaign per region
  one ad set per month of winners,
  per Facebook page
    winner fades ........... GRAVEYARD

GRAVEYARD
  bid cap, all markets, one campaign

Testing finds winners. Winners scales them, split by region. Graveyard keeps two kinds of ad alive on a bid cap: good ads that couldn't scale, and old winners that have faded.

Testing campaign: ABO with a target ROAS or target CPA bid strategy

The testing campaign uses ABO, Meta's ad set budget optimization, so every ad set gets its own budget.

• One testing campaign per product. Products with different margins need different target CPAs. Put them in one campaign and Meta will chase the cheapest CPA, even when that product makes you less.
• The bid strategy is a target ROAS (tROAS) or a target cost per result (tCPA), whichever matches the brand's main KPI. Set it from break-even.
• One ad set per creative batch, with 3 to 5 ads in it.
• Each ad uses Meta's multiple text options, with 3 primary texts, 3 headlines and 1 description.
• Targeting is the brand's top-performing countries, with no interests. Let the creative find the buyer.
• Each batch is built as a new ad set, switched off with its ads switched on, so one toggle launches it and nothing else in the account moves.

How we fill those batches every week, and how we read them, is in our creative testing framework.

There's no fixed test budget per ad. The bid target controls spend, and Meta's delivery decides which ads deserve it. Lots of ads getting little or no spend is expected, and it's no reason to test fewer.

If other agencies run tests in the same account, you can split testing into one campaign per product per agency, so everyone can see their own tests. Splitting it that way doesn't change performance.

Where each test ad goes

A winner is an ad that takes real spend and keeps taking it under the account's cost controls. A one-day spike doesn't count. A high-ROAS ad that can't spend doesn't count either.

Winners campaigns: bid cap with an inflated budget

• The bid strategy is bid cap.
• The budget is set well above what you expect to spend. The cap controls spend, so the campaign only spends when it can buy results at or under the cap.
• One campaign per region: US, UK, CA, AU, NZ and SG, plus any other market the brand sells into.
• One ad set per month of winners, split by the Facebook page the ads run from.
• New winners go into the current month's ad sets. Earlier months stay live for as long as they keep performing.

For a brand running ads from three pages, each region campaign looks like this:

Why the budget is set high on purpose

It comes down to one rule. If an ad or an account can't spend profitably, the offer, the landing page or the creative isn't good enough yet. When ROAS drops as spend rises, the account is telling you which of those to fix. We don't treat it as a ceiling, and we don't plan around per-ad spend caps.

With a bid cap and a big budget, the campaign spends wherever it can buy results at or under the cap and nowhere else. If you're stuck with the opposite problem, an account that won't spend at all, this is the checklist I'd start with. For the difference between the two capped bid strategies, see cost cap vs bid cap.

Graveyard campaign: where good ads go instead of being switched off

Graveyard is one bid-cap campaign across all markets. Two kinds of ad go in:

• Test ads with good signals that weren't strong enough to scale.
• Past winners that have faded out of their region's Winners campaign.

When a winner fades, it moves here. We don't just switch it off. On a bid cap it only spends when it can still hit the cap. If Graveyard spend suddenly jumps, tighten the cap.

Some accounts we inherit already run a "Losers" campaign. It's the same idea under a different name.

Settings shared by all three campaigns

Every campaign in this structure is for new-customer acquisition. If a brand doesn't have one of those exclusion audiences built yet, we build it before launch. There's a full walkthrough in our post on exclusion audiences, and the attribution windows are explained in what 7-day click attribution means.

One naming pattern

One pattern across every account lets you report Testing, Winners and Graveyard spend without tagging anything by hand.

So a testing ad set might read 2026-09-30 | Batch 14 | US-UK-AU.

The routine that keeps it running

1. Daily: check every live test ad and switch off the ones going nowhere. Move any ad that hits the winner rule into this month's Winners ad set in each region campaign.
2. Weekly: check each Winners campaign for winners that have faded, and move them to Graveyard.
3. Monthly: on the 1st, open the new month's Winners ad sets, one per page, in every region campaign.

CBO or ABO: what changed since 2025, and why

In my May 2025 video on scaling Meta ads, the setup was one CBO (campaign budget optimization) campaign for testing with minimum spends, plus an Advantage+ shopping campaign on a cost cap for scaling. Winners got pushed from one to the other. As late as January 2026 I was telling people to put all their ads in one CBO and move the budget up and down. The CBO vs ABO post covers how the two budget types work.

Four things changed:

1. Testing moved to ABO with one ad set per batch. Each week's batch gets its own budget and its own switch, so it launches as a unit and you can see how that batch did on its own.
2. Bid targets replaced minimum spends. On a tROAS or a bid cap, Meta decides which test ads earn money, so there's no per-ad budget to manage.
3. Scaling moved to bid-cap Winners campaigns split by region and by month. The cap controls spend, so the budget can sit well above what we expect to use.
4. Faded winners and near-misses now have a campaign of their own in Graveyard.

Plenty stayed the same. We still go broad, keep the campaign count low, give each margin its own campaign, and treat creative as the main lever. Like every SOP we run, this one is open to challenge. If you can beat it, show us the numbers and we'll change it.

FAQ

CBO or ABO: which is better for testing Meta ads?

We test on ABO with a target ROAS or target CPA, one ad set per creative batch. I used a single CBO for testing in 2025. The ABO setup gives each batch its own budget and its own on/off switch, and the bid target decides which ads get spend.

Which bid strategy should a scaling campaign use?

Our Winners campaigns use a bid cap with a budget set well above expected spend, so the campaign only spends when it can buy results at or under the cap. Testing runs on tROAS or tCPA.

What is a Graveyard campaign in Meta ads?

One bid-cap campaign, across all markets, for test ads that showed good signals but couldn't scale and for winners that have faded. It only spends when those ads can still hit the cap.

Do you use 7-day click, 1-day view attribution?

We use 7-day click with 1-day engaged-view on every campaign in this structure.

How many ads should go in a testing ad set?

3 to 5, with one ad set per creative batch. Each ad carries 3 primary texts, 3 headlines and 1 description.