Meta's AI creative model just changed. Your June tests are already stale.

You ran creative tests in June. You found a winner. You scaled it. That's good practice. Here's the problem: on July 7, Meta announced the Muse Image model and began rolling it into Advantage+ Creative across Q3. The generation stack underneath your ad variations changed. The quality baseline shifted. Your June opt-out decision might be wrong now.

This isn't a minor update. Muse replaces the model that was producing the quality issues that made a lot of brands turn off Advantage+ Creative enhancements in the first place. If you're one of them, you need to re-test.

Advantage+ Creative reaches 8 million advertisers. Campaigns on it average 12% higher CTR than manual alternatives, per Meta Marketing Science 2025. If your agency still hasn't re-enabled it after Q2 quality concerns, they're running a stale playbook on a tool that just had a material upgrade.

What the Muse Image model actually changes for Meta Advantage+ Creative

Advantage+ Creative takes your uploaded assets and automatically generates variations: different backgrounds, formats, text overlays, and crops. Meta's delivery algorithm then tests those variations to find the combination that converts best for each user segment.

The problem was that the AI model generating those variations was producing low-quality outputs. Images that looked clearly AI-generated. Blurry backgrounds. Product rendering issues that made the creative look unfinished. Understandably, a lot of DTC brands turned it off rather than let Meta serve creative that made the brand look bad.

Muse is the new generation model. Meta announced it July 7 with planned rollout into Advantage+ Creative through Q3. It's built on a different architecture than what was running before. The quality issues that justified opting out in June were tied to the old model. They're not guaranteed to carry over to Muse.

One thing worth knowing before you re-enable it: Meta still doesn't give you preview access to AI-generated variations before they go live. You'll see them in the ad reporting view after delivery. If your brand has strict visual standards, monitor the first 48 hours closely. Any variation that doesn't pass your quality bar can be turned off individually without killing the whole ad set.

Meta Push Delivery Control: the budget cannibalization fix

Budget cannibalization has been one of the worst problems in Meta ads for years. One incumbent ad finds an audience, starts converting, and eats the entire ad set budget. Newer creatives never get enough impressions to reach statistical significance. Your creative testing process breaks before it starts.

Meta's answer is Push Delivery Control, which rolled out to nearly all accounts by July 10, 2026. You pick an ad, and Meta forces 20-30% of your ad set budget to that creative for up to 7 days. It overrides the algorithm's natural preference to serve what it already knows converts.

This matters most if you run a test-and-scale creative strategy. Without Push Delivery, a new creative against a proven winner never gets enough impressions to prove itself. Push solves that by guaranteeing budget to the challenger. Real data on new creatives in 48-72 hours instead of waiting two weeks and getting no clarity.

I run Push Delivery on every new creative for client accounts when a proven winner is already live. Without it, new ads starve. With it, you get usable signal before the week is out. For the full picture on how AI marketing for ecommerce handles creative testing at scale, the budget allocation layer is where most of the leverage lives.

The CPM context you need

Neither of these tools runs in a vacuum. Median Facebook CPM in June 2026 hit $21.90, up 16.5% year-over-year, according to AdMake AI's July 2026 platform analysis. Every thousand impressions costs more. Creative efficiency isn't optional anymore when CPMs keep climbing.

That's why the Muse upgrade matters beyond the quality question. If Advantage+ Creative generates 12% higher CTR on your existing assets at no additional production cost, that's direct CPM offset. The same budget gets more clicks at lower effective cost per visit.

Top-performing DTC brands already achieve 35-45% lower CAC than industry averages. That gap doesn't come from bigger budgets. It comes from getting more out of each dollar through smarter creative selection and higher testing velocity. Muse and Push Delivery are both tools in that stack.

What to actually do this week

Three things, in order.

Check your Advantage+ Creative settings.If you turned off AI-generated variations in Q2, go back in and re-enable them on one ad set. Not all of them. One. Let it run for a week and look at the actual variations Meta serves. Muse is what's generating those now. If the quality holds, expand from there.

Set up Push Delivery on one challenger creative. Find a new ad that hasn't had enough impressions to prove itself. Push 20-30% of that ad set budget to it for 7 days. You get real signal before the week is out. If it wins, scale it. If it loses, you know quickly and move on.

Raise your creative testing volume baseline.Beauty brands benchmark 15-20 new ad tests per month. Most verticals can't match that, but the direction matters. More tests produce more winners. Tools like AI-generated creative make hitting higher volumes possible without production budget. And if you're running AI creative pre-scoring, you can filter out likely losers before they spend a dollar.

The brands that fall behind on Meta aren't ignoring it. They're running it on settings they configured six months ago and treating the results as the market. Two material tool releases in one month changes what those settings should be.

What running this on client accounts actually looks like

At Venti Scale, every client account on Meta gets a standing audit of which controls are enabled. When Meta ships something like Push Delivery Control, it goes into the account review cycle within the same week, not the next quarterly check-in.

Most agency retainers run on quarterly review cycles. The account manager checks in. The report goes out. The settings stay where they were. That model made sense when platform updates shipped on quarterly schedules. It doesn't make sense when Meta ships meaningful changes in the same week it announces them.

If your agency's response to "did you see the Muse Image rollout?" is a blank look or a promise to include it in next month's strategy call, that's not a personality problem. It's a structural one. Monthly retainer plus quarterly review can't keep up with a platform that ships on a weekly cycle. The brands winning on Meta right now are the ones treating platform updates like operational tasks, not content to read later.

Dustin Gilmour

Founder of Venti Scale. I manage Meta ad campaigns for ecommerce brands and review Advantage+ Creative performance every week across active client accounts. The creative testing observations here come from live accounts, not vendor benchmarks.

About·LinkedIn·X·Updated July 21, 2026