The oldest frustration in Meta Ads is one they spent years ignoring. You launch a new creative. The existing ad — the one that ran strong three months ago — eats the vast majority of delivery. Your new asset sits at zero spend for days and you never get a real read on whether it works.
The workaround has always been the same: duplicate the ad set, isolate the new creative, run it in parallel, then consolidate or shut it down after a week. It worked. It also fragmented your pixel signal, reset the learning phase in two places simultaneously, and required more cleanup than the test was worth.
Meta just fixed this. The feature is called Push Delivery to This Ad, and practitioners reporting on the rollout confirm it reached nearly all accounts by July 10, 2026.
What the feature actually does
Push Delivery is a manual delivery override at the individual ad level. Inside an existing ad set, you select one specific ad, set a percentage of the ad set's daily budget to allocate toward it, and choose a duration — up to seven days.
The percentage and window are both set by you. Once the window closes, Meta's delivery system resumes its normal algorithmic split. The other ads in the ad set keep running throughout; they just receive a smaller budget share while the push is active.
The feature requires at least two ads in the ad set to appear. With one ad live, the option isn't shown. Add a second ad and it becomes available under the extended settings.
Meta has not published an official help page for this yet. The practitioners who reported on it — Jon Loomer, Foxwell Digital — describe consistent behavior across accounts, but treat the specific mechanics as provisional until Meta documents it formally.
Why this changes creative testing
The standard creative testing workflow has been either patient or expensive. Either you drop a new ad into a live ad set and hope the algorithm picks it up (it usually doesn't if an incumbent is already established), or you duplicate the ad set to force spend isolation, which means splitting your audience data and managing two sets of learning phases.
Push Delivery removes that tradeoff. You can launch a new video or static into your best-performing ad set, push a portion of the daily budget toward it for five days, and get a real performance read without pulling delivery from your winning creative entirely. After the window closes, Meta takes back full algorithmic control. If the new ad earned its place, it competes normally. If it underperformed, you have data — not a theory built on zero impressions.
For brands on a two-to-four week creative refresh cycle, this is a genuine process improvement. The ad set duplication workaround worked, but it added overhead and introduced measurement noise that made comparisons harder. This keeps everything inside one ad set, one learning phase, one pixel signal pool.
When to use it and when not to
Push Delivery makes sense in four situations: launching fresh creative into a campaign where one incumbent ad dominates delivery; running time-boxed promotions where you need a specific asset live during a sale window; testing a creative concept before committing to a dedicated ad set; and refreshing a fatiguing top ad without disturbing the account's overall learning state.
It's less suited to: new campaigns where the algorithm hasn't established a delivery baseline (the learning phase effect is more disruptive early on); accounts with only one ad per set (the option simply won't appear); and situations where you need a controlled A/B comparison with audience isolation. For structured head-to-head tests with isolated variables, Meta's native split testing tool still produces cleaner data.
One thing to be clear about: pushing delivery to an ad guarantees it spends. It does not guarantee it performs. If the creative is weak, you'll spend money learning that faster than you would have otherwise. That's the point — faster learning cycles, not improved creative quality.
What Meta still hasn't returned
One useful control doesn't reverse the trajectory. Meta has methodically removed advertiser levers over the past five years — granular placement controls, audience targeting precision, creative variation management. Push Delivery is a real addition, but it's one lever in an account where most others have been automated away.
There's still no way to allocate budget by placement, no way to prevent Advantage+ from modifying your creative without turning off the feature entirely, and no audience-level budget control that isn't mediated by the algorithm. The fact that Push Delivery still doesn't have a published help page is a signal about where it sits on Meta's priority list relative to its AI-driven automation roadmap.
Use it. Stop duplicating ad sets to force creative tests. But don't confuse a single feature restoration with a change in direction.
If you want to see which Meta campaigns are actually driving growth versus which are spending on autopilot, the free audit at gromerce.com/audit shows the breakdown in under three minutes.
The algorithm always preferred your old winner. Now you can briefly override it.
Sources: Jon Loomer Digital, Foxwell Digital, July 2026
