The promise versus the reality
Meta’s Advantage+ campaigns have gone from experimental to default in most ad accounts. The promise is simple: hand Meta your creative, your product catalogue, and your budget — the algorithm handles targeting, placements, and optimisation. In many cases, it delivers. In others, it silently bleeds spend on low-intent placements while the dashboard shows a healthy-looking ROAS.
The difference between the two outcomes is rarely the algorithm. It’s almost always the inputs.
What Advantage+ does well
For e-commerce accounts with a populated catalogue, decent pixel data, and at least three to four creative variants, Advantage+ Shopping consistently beats hand-built campaigns on cost per purchase. The system has access to signal you don’t — cross-account behaviour, device-level patterns, time-of-day intent — and it churns through audience permutations faster than any media buyer can.
For app install accounts, Advantage+ App Campaigns are now table stakes. If you’re still running manual app campaigns with hard-coded interest targeting, you’re leaving installs on the table.
Where it quietly burns money
Three failure modes show up repeatedly in audits:
Creative starvation. Advantage+ needs at least four to six distinct creatives to find combinations. Feed it two static images and it’ll spend on whichever one wins early — even if that early signal is statistical noise.
Pixel pollution. If your Purchase event is firing on add-to-cart, or your Lead event is double-counting, Advantage+ optimises to the wrong outcome. The campaign looks profitable on Meta’s dashboard and unprofitable in your CRM. We see this on roughly one in three accounts we audit.
Audience overlap with manual campaigns. If you’re running Advantage+ alongside hand-built lookalikes, you’ll cannibalise. The auction sees both campaigns bidding on the same person, and your CPMs drift up.
The override framework
We use a simple test before letting Advantage+ run unsupervised:
- Are conversion events clean? Run the Meta Pixel Helper on every key page. If a Purchase fires twice or a Lead fires before form submit, fix it before launching.
- Is CAPI live and deduplicating? Server-side events should match browser-side events with matching event IDs. Without this, Advantage+ optimises to incomplete data.
- Do you have at least four distinct creative angles? Different hooks, different value props — not four edits of the same video.
- Is your daily budget at least 50× your target CPA? Below that, the algorithm doesn’t have enough room to learn.
If all four are yes, let Advantage+ run. If any are no, fix them first or stick to manual.
The metric that actually matters
Don’t judge Advantage+ on Meta’s reported ROAS. Compare it against your CRM revenue or shop backend. The gap between the two is your attribution discrepancy — and it’s usually the truest indicator of whether the algorithm is working or just looking good.