Advantage+ ate your Meta strategy.
Something strange happened to Meta Ads in the last 18 months. The job of the media buyer basically inverted.
In 2022, a good Meta marketer was someone who could build a 47-ad-set campaign structure, stack lookalike audiences, layer interest targeting, and manually pause the bottom 20% of ads every Tuesday. That was the job.
In 2026, all of that is actively making your campaigns worse. Meta's Advantage+ Shopping (now rebranded Advantage+ Sales) plus the Andromeda algorithm underneath it now runs prospecting, retargeting, audience selection, placement optimization, and creative rotation — all in a single consolidated campaign. And the research is pretty unambiguous about this: Advantage+ campaigns deliver $4.52 in revenue per $1 spent, a 22% higher return than manually structured campaigns. Meta's own data shows a 10% CPA reduction; agencies with clean setups are seeing 15-25% improvements.
The algorithm got good. Really good. So the skill shifted.
What the algorithm now owns
Before we talk about what matters, let's be clear about what doesn't anymore:
Audience targeting. No lookalikes. No interest stacks. No saved audiences. Meta decides who to show your ad to, using signals from your pixel + CAPI that are 10x richer than anything you could define manually. Your five interest groups you were so proud of in 2022 are a rounding error in its targeting math now.
Placement selection. Feed, Reels, Stories, Marketplace, Messenger, Audience Network — Advantage+ Placements decides where each impression serves. Trying to force-select placements almost always costs you efficiency.
Budget allocation. Campaign Budget Optimization shifts spend between prospecting and retargeting, between creatives, between times of day. You can set an existing-customer cap (more on that below) but the rest is algorithmic.
Ad rotation and testing. You upload up to 150 creative variants; the algorithm tests and kills losers automatically. Manual A/B tests with statistical significance thresholds — the thing we all did for a decade — is now a way to slow the algorithm down, not speed it up.
In 2026, your competitive edge on Meta isn't in outsmarting the algorithm. It's in feeding it better inputs than your competitors do.
What still matters — the five levers
Here's where actual work happens now. These are the inputs that determine whether Advantage+ prints money or burns it.
1. Creative quality and volume
This is the whole ballgame. Since you don't control audience targeting, the algorithm uses creative signals to identify who to serve the ad to. Better creative means better audience matching means lower CPA. The optimal mix:
30-40% UGC and testimonials (lowest-CPA format in e-commerce). 20-30% product demonstrations. 15-25% lifestyle content. 10-20% promotional / offer-driven. Mix static and video.
More importantly: volume matters. Brands running 3-5 unique creative angles per campaign with 15+ variants at all times are consistently outperforming brands running 5 total. The algorithm needs raw material to optimize with. If you're uploading 3 ads and leaving them for 6 weeks, you're leaving money on the table regardless of how good those 3 ads are.
2. Pixel + Conversions API (both, not one)
iOS and ad blockers eat browser-based pixel events. Server-side CAPI catches most of what the pixel misses. Running both gives Meta's algorithm ~2-3x more conversion signal than pixel alone — which is the single largest determinant of whether your campaign escapes the learning phase.
Speaking of which: 50 conversion events per week is the minimum to fully exit learning. If your campaign isn't getting there on current budget, either consolidate campaigns or raise budget until it does. A campaign stuck in learning will spend 20-50% more per conversion indefinitely.
3. The existing-customer cap
The single most-fumbled setting in Advantage+. By default, the algorithm will over-index on retargeting because existing customers convert more easily. Result: inflated ROAS numbers, but very little actual new-customer acquisition.
Set your existing-customer cap based on your actual business goal:
0% — pure prospecting. Use this when you're scaling aggressively and need net-new growth.
10-20% — mostly prospecting with a light retargeting assist. Sensible default for most growing brands.
30-50% — hybrid mode. Use when you have a mature customer base and retention economics are strong.
Most brands I audit are unintentionally running at 40-60% retargeting. The ROAS looks great and the actual business is slowly starving for new customers.
4. Campaign consolidation
The old "more campaigns = more granular control" instinct is wrong now. Every additional campaign you run fragments your budget and denies the algorithm the conversion volume it needs to learn. Most accounts should be running 1-2 Advantage+ campaigns, max. One for core catalog, maybe a second for a specific product line or seasonal push.
If you currently have 8 active campaigns, three are probably in perpetual learning phase. Consolidate.
5. Budget floor
Meta's math wants $50/day minimum to exit learning within 7 days. For accounts with target CPAs above $50, you need more — $100-200/day per campaign is typical for anything with a meaningful ticket price. Underfunded Advantage+ campaigns produce underwhelming results and get killed before they'd have worked. Spend enough to give the algorithm real data, or don't bother.
What to stop doing
A "stop doing" list is often more useful than a best-practices list. For Advantage+:
Stop running manual prospecting + retargeting as separate campaigns. Advantage+ does both better than your split structure does, with the added benefit of shared conversion data between them.
Stop narrowing age ranges. Unless you have legal reasons (alcohol, firearms, finance), broad targeting outperforms narrow. Let the algorithm decide.
Stop pausing the lowest-performing ads manually. The algorithm is already doing that. Manual intervention resets signals and restarts learning.
Stop doing A/B tests as your primary optimization method. Run experiments at the campaign or offer level (different landing pages, different messaging) — not at the ad level within a single Advantage+ campaign.
Stop believing your last-click attribution. Meta's Incremental Attribution (launched late 2025) measures view-through and upper-funnel lift. If you're only counting last-click, you're undervaluing Meta by 30-50% and over-rotating to channels that are easier to measure.
The new job description
If you manage Meta Ads at a brand or agency, your skill set has to change. The job now is:
60% creative strategy — briefing, producing, and curating a deep and diverse creative library.
20% technical foundation — pixel, CAPI, catalog, feed health, conversion events.
15% measurement — incrementality tests, MMM inputs, post-purchase surveys that tell you what attribution can't.
5% campaign management — because honestly that's all it should be now.
If you're still spending 60% of your time in Ads Manager tweaking audience settings, you're doing 2022's job in 2026's platform.
If your Meta campaigns feel like they're "working" but you can't tell whether Advantage+ is actually set up right — or if you're stuck spending in learning phase — book a free 30-minute audit. I'll pull up your account with you and walk through exactly what's working, what's fighting the algorithm, and what to change first.
