Which Advantage+ Creative Enhancements to Turn Off in 2026

Many accounts run Advantage+ creative enhancements without ever deciding to. They ship on by default, they quietly rewrite your copy, animate your stills, and paint generative pixels around your product, and they do it in the name of “performance.” Some of that genuinely helps delivery. The rest buys you cheaper, lower-intent impressions that flatter your reported numbers while dragging real contribution margin. The job is to tell the two apart.

This is a keep-or-kill list. For each enhancement, the question is the same: does it protect brand fidelity and margin, or does it just inflate cheap reach? Decide it deliberately, then turn off what fails.

For the surrounding account decisions, compare Advantage+ Full Funnel: Where Automation Helps vs Burns Budget and use Meta Made Advantage+ the 2026 Default: What Changed as the next diagnostic.

The test that decides keep or kill

Two filters, in order.

1. Fidelity. Does the enhancement change what you actually said or showed? Anything that rewrites your claim, alters your product’s appearance, or attaches words you didn’t write is a fidelity risk. You are legally and commercially responsible for copy you never approved.
2. Margin, not platform ROAS. Many enhancements work by widening your creative into more (and cheaper) placements and impressions. Reported ROAS can hold or even tick up while CPA-to-margin quietly worsens, because the incremental traffic is lower-intent. Judge every enhancement on contribution after COGS, shipping, and fees, not on the in-platform number.

If something fails fidelity, kill it. If it passes fidelity but you can’t prove it on margin, it goes in the test pile, isolated, not left on by faith.

A practical note: you no longer have to accept the whole bundle. The enhancements toggle individually under the creative edit panel, so “all on” is a choice, not a default you’re stuck with.

Kill these by default

Text improvements (auto-rewriting your primary text and headline). This is the worst offender. The system rephrases your copy “to perform better,” which means your specific claim, your offer mechanics, and your voice get smoothed into generic ad-speak. You lose the exact promise you tested, and you can end up running statements you’d never sign off on. Turn it off. Write your own copy and keep it.

Music. Auto-added audio on a static or product-led ad seldom lifts intent. It pushes your asset toward sound-on, scroll-heavy placements where impressions are abundant and cheap and buying intent is thin. It also flattens brand fidelity, because the track is chosen for you. For most catalog and offer ads, kill it.

Image animation (pan/zoom and faux-3D motion on stills). The Ken Burns drift and pseudo-parallax effects exist mainly to make a still eligible for motion placements. That’s the tell: you’re buying placement breadth, not persuasion. It can also distort a clean product shot. Unless you’ve shot for motion, this is impression inflation. Kill it and test real motion creative separately.

Image expansion and background generation (generative outpainting and swapped backgrounds). Generative fill invents pixels around your product to fit new aspect ratios. On lifestyle scenes it’s sometimes survivable; on a product hero it’s a fidelity hazard, the model can warp geometry, add phantom edges, or fabricate a setting that misrepresents the item. Misrepresenting a physical product is a returns-and-disputes problem, not a creative one. Kill it for product, and crop deliberately instead.

Relevant comments (surfacing social comments under the ad). You don’t control which comments get elevated, and a single skeptical or off-topic one sits next to your offer at the moment of decision. The upside is marginal; the downside is uncapped. Kill it.

Keep these on

Visual touch-ups (brightness, contrast, light sharpening). Low-risk, non-destructive, and they don’t change your claim or your product’s identity. They mostly make under-lit user-generated footage usable. Keep on, spot-check that they aren’t over-saturating brand colors.

Cropping and aspect-ratio adjustments. You need your creative to fit feed, story, and reel frames cleanly. Letting the system fit existing pixels into each frame is fine, that’s framing, not fabrication. Keep on, but pre-build key ratios for hero assets so the auto-crop never has to guess on your most important creative.

Catalog and product tags. Surfacing the actual SKU, accurate price, and a tap-to-product path raises intent and shortens the path to purchase, the opposite of cheap-impression inflation, provided your catalog feed is clean. Keep on, and audit the feed, because a wrong price here is a real fidelity and margin leak.

Site links / additional destinations. Extra navigational links to real, high-intent pages help qualified users self-route. They don’t touch your creative or your claims. Keep on.

Test these, then judge on margin

These aren’t villains, but they’re not free passes either. Isolate each one, run it against a clean control, and read contribution, not platform ROAS.

Text generation (machine-written copy variants). This can be a useful idea generator, and occasionally a variant beats yours. But left fully automatic, it ships language you didn’t vet. Treat it as a draft source you approve, never as an always-on autopilot. Test it; keep only the lines that win on margin and that you’d actually stand behind.

Overlays and info labels (price badges, shipping or promo stickers). A genuine, accurate incentive overlay can lift intent. A cluttered stack of auto-applied badges cheapens a premium product and trains buyers to wait for a deal, which compresses margin over time. Test one clean, true overlay at a time; kill the pile.

Enhance CTA and minor layout nudges. In many cases harmless, occasionally helpful. Low priority, but still worth confirming the CTA the system lands on matches the action you actually want, rather than the highest-click-rate one.

How to read the results

The trap with every one of these is that the platform grades them on its own scorecard. More placements, more impressions, more clicks, all reported back as success. So before you trust any “keep,” check the unit economics underneath:

• Is CPA staying under your margin ceiling, or just under your old CPA?
• Did frequency climb without a matching lift in purchases? That’s reach you paid for that isn’t converting.
• Did the cheaper impressions dilute intent (rising click volume, falling conversion rate)?

As an illustrative planning frame, it’s common for a meaningful slice of spend, think roughly a fifth to two-fifths in a loosely-governed account, to sit in low-intent inventory that looks fine on platform metrics and poor on margin. That’s a range to investigate, not a fixed number. A read-only operator like Bach can flag when an enhancement is silently rewriting a headline or pushing you into placements that don’t pay back, and surface it for your approval before anything changes, but the decision and the margin math stay yours.

The takeaway

Default-off the enhancements that alter your words or your product, text rewrites, music, faux-motion, generative backgrounds and expansion, and surfaced comments. Keep the ones that are non-destructive or genuinely raise intent, touch-ups, sensible cropping, accurate catalog tags, real site links. Everything else gets isolated and judged on contribution margin, never on the in-platform ROAS. Deciding which Advantage+ creative enhancements to turn off isn’t a creative preference; it’s the difference between buying intent and buying impressions.