Meta Made Advantage+ the 2026 Default: What Changed

If you opened the campaign builder recently and felt like the controls moved, you weren’t imagining it. The shift to Advantage+ as the default path isn’t a new ad product so much as a quiet renegotiation of who decides what. Targeting you used to set as a hard rule is now a hint. Reporting you used to read at a glance is now one layer further from you. Most of this happened by default, with a toggle most operators never flipped back.

This is a plain-English account of what actually changed: which manual controls survived, which got softened into suggestions, and which simply left the building.

For the surrounding account decisions, compare Which Advantage+ Creative Enhancements to Turn Off in 2026 and use Creative Testing Inside Advantage+: How A/B Testing Changed as the next diagnostic.

The real change: a default, not a new feature

For much of the last few years, Advantage+ campaigns and “manual” campaigns lived side by side as separate things you chose between. The 2026 consolidation collapsed that. Now there is essentially one sales (and app) campaign flow, and Advantage+ is switched on inside it by default. You can turn it off and run the campaign closer to manual, but the burden of proof flipped: automation is the assumption, and manual is the opt-out you have to go find.

That sounds cosmetic. It isn’t. When Meta made Advantage+ the 2026 default, it changed what happens when someone builds a campaign quickly and doesn’t audit every setting — which is most campaigns, much of the time. The path of least resistance now hands more decisions to the system.

What survived (and is still worth setting deliberately)

The good news: the controls that actually protect your unit economics are mostly intact. Use them on purpose.

• Cost controls. Cost-per-result goals, ROAS goals, and bid caps still exist and still bind. This is your real steering wheel. If you care about hitting a target CPA-to-margin ratio, this is where you enforce it — automation does not override a goal you set.
• The existing-customer budget cap. You can still tell Meta what share of a campaign’s budget is allowed to chase people who already bought. This survived for a reason: it’s the one lever that keeps a “blended” campaign from quietly spending most of its budget retargeting your warmest list and reporting a flattering ROAS that’s really just harvest.
• Manual placements. Still available, just buried under the Advantage+ placements default. If you have a genuine reason to exclude a surface, you can.
• Campaign-level budget and structure. Consolidated, fewer-ad-set structures are encouraged, but you still control budgets, schedules, and creative.
• Exclusions in most flows. You can broadly still exclude your customer list, recent purchasers, or specific audiences — important for keeping prospecting honest.

If you treat those five as your non-negotiables on every build, you keep the parts of the account that move your margin.

What got softened into “suggestions”

This is the part operators underestimate. Several controls didn’t disappear — they were demoted from rules to signals.

The clearest example is audience definition. The interests, demographics, and lookalikes you enter increasingly function as a suggestion the system can spend outside of, rather than a fence it must stay inside. Meta’s framing is that it uses your input as a starting signal and then finds conversions wherever the model expects them. In practice, that means “25–34, interested in skincare” is now closer to a hint than a constraint, and your spend can drift well beyond it if the model thinks it’ll perform.

Detailed-targeting expansion follows the same logic — on some objectives it’s effectively always on. The mental model to adopt: you are no longer drawing the audience. You are seeding it and capping the downside. Your real targeting controls now are your creative (which selects an audience implicitly), your exclusions, your cost goal, and your existing-customer cap. Everything you type into the interest box is advisory.

What quietly disappeared

A few things didn’t survive the consolidation, and their absence is easy to miss because nothing errors out — you just lose resolution.

• Granular audience reporting. Because the audience is now broad-and-blended by design, the breakdown of which segment drove a result is thinner. You can still see placement and some demographic breakdowns, but the clean “this lookalike vs. that interest stack” comparison that justified a lot of old optimization work is largely gone.
• Clean prospecting/retargeting separation by default. When new-customer and existing-customer spend live in one campaign, the old habit of reading two separate campaigns to judge true new-customer efficiency stops working. The existing-customer cap is now your only real instrument for that split — and only if you set it.
• Some manual structural choices, folded into the recommended consolidated setup. You can frequently rebuild them, but they’re no longer the default scaffolding.

The visibility tax you’re absorbing

Here’s the tradeoff almost nobody priced in. The 2026 default is genuinely good at finding conversions. What it’s worse at is telling you why. You get the result; you get less of the reasoning.

That has three downstream effects on how you should operate:

1. Platform ROAS gets less trustworthy as a diagnostic. A blended campaign can post a strong return that’s mostly retargeting and remarketing your existing base. Without the existing-customer cap and without segment-level clarity, you can’t easily tell incremental growth from harvesting. Judge the account on MER and contribution margin, not the in-platform number, and watch your new-customer share like a hawk.

2. Edits cost more than they used to. Broad, consolidated, automation-led campaigns lean hard on the learning phase. Every significant edit can re-trigger it, and with less granular feedback you’re flying with fewer instruments while it re-stabilizes. As a planning rule of thumb — not a Meta-published assurance — assume a campaign needs a meaningful run of recent optimization events before its numbers mean anything, and resist judging or editing inside that window. Meta needs enough recent signal to optimize; thrash starves it.

3. Diagnosis moves off-platform. Because the interface shows you less of the how, the work of understanding wasted spend, audience overlap, and frequency creep shifts to your own reading of the data. This is exactly the gap Bach AI was built to sit in: it reads the account continuously, separates harvest from incremental growth, flags where the default is quietly spending against your margin, and surfaces the change to make. It stays read-only until you approve — Meta changes only execute on your say-so, and on the Google side it’s intelligence only, no execution.

The practical takeaway

Treat the 2026 default as a competent junior buyer with a habit of optimizing for the most straightforward conversions, not the most valuable ones. Your job moved from drawing the audience to setting the guardrails and reading the truth.

Concretely, on every build:

• Set a real cost or ROAS goal tied to your margin, not a vanity target.
• Set the existing-customer budget cap on purpose, every time — it’s your only clean new-customer instrument now.
• Keep exclusions tight so prospecting stays prospecting.
• Judge performance on MER and new-customer contribution, never the blended in-platform ROAS alone.
• Leave campaigns alone long enough to stabilize before you trust or touch them.

The default didn’t take away your control. It moved it — from the targeting box you used to fuss over to the guardrails and the reading of results you now can’t skip. Operators who notice the move keep their margin. Operators who don’t quietly hand it over.