I've seen many accounts move everything into Advantage+ on their Meta rep's advice, then declare three weeks later that "results got worse, automation doesn't work." The interesting part: over the same period I've watched accounts scale seriously with Advantage+. The difference wasn't the campaign type; it was what the campaign was fed.
What automation takes over — and what it can't
Advantage+ sales campaigns take targeting, placement and budget-allocation decisions away from you. But there are three things they can never take over: signal (the quality of conversion data flowing from pixel/CAPI), creative (the variety and angles of the assets you supply) and economics (knowing which conversion actually makes money). Most accounts switching to automation leave all three in their old, unmaintained state. The result: the system runs faster with a broken compass.
Mistake 1: automation on dirty signal
In the post-iOS era, on accounts with incomplete CAPI, double counting, and no refund/cancellation data, what Advantage+ "learns" isn't real sales — it's a corrupted proxy. The system grows the audience that appears to convert, even if that audience returns products, fails payment, or hunts one-off discounts. The question to ask before automating isn't a campaign setting: "How closely does the conversion data flowing into the system match my accounting?" If the match is below ~85%, fix measurement first, automate second.
Mistake 2: expecting the machine to learn from three creatives
Advantage+'s real power is matching many creative variants to different micro-audiences. In a campaign launched with three images there is nothing to match; the system piles onto the best performer, and when that creative fatigues, the campaign collapses with it. My working standard: at least 8–10 meaningfully different angles per campaign — problem-led, social-proof-led, price-led, use-case-led — with a weekly creative refresh rhythm. Automation doesn't reduce creative work; it turns creative into the single biggest lever you have left.
Mistake 3: surrendering without guardrails
"Advantage+ is a black box, you control nothing" isn't true either. Existing-customer budget caps, country exclusions, creative-level reporting, placement breakdowns — the controls exist; they just go unused. The most critical one: leaving the existing-customer limit empty while believing you're acquiring new customers. The system goes wherever conversion is easiest, and that place is usually the audience that would have bought from you anyway. ROAS shines in the report; there are no new customers in the revenue.
When Advantage+, when manual?
My practical frame: with clean signal, sufficient creative volume and a daily budget large enough for the system to learn, Advantage+ usually beats a manual setup. For niche B2B audiences, very narrow geographies, brand campaigns needing strict frequency control, and new accounts with weak signal, manual structures still win. The decision isn't ideological; it's conditional. "Is automation good or bad" is the wrong question — the right one is: "Does this account meet the conditions under which automation works well?"
In short: Advantage+ is a multiplier. On a well-fed account it multiplies good outcomes; on a neglected one it multiplies the mistakes. The job isn't to step aside — it's to move your work up into the signal-creative-economics layer. I covered how I build the Meta signal side in the CAPI article.