What I noticed this week
I've been doing setups in Meta Ads Manager for a few years now. Sometimes Meta changes something and it's felt right away. Advantage+ Audiences was such a moment, it flipped our setup logic completely. Sometimes it's a UI shift that seems harmless but if you don't pay attention, you build a setup that can't be edited cleanly later.
Two things from that second category this week. Both communicated quietly. Both with consequences that turn annoying in two weeks if you don't classify them now.
Update 1: Placements moved to ad level
In the new Meta Ads UI rolling out to many accounts since early May, you can upload multiple images and videos in a single ad. That's nice, long overdue.
But with that comes a subtle shift: placements are now picked at ad level, not just ad set level. Sounds convenient at first. Per ad you can decide which placements each creative serves. Format-specific, creative-optimized. The 9:16 video gets the Reels placement, the square image gets Feed plus Stories, and so on.
The placement-edit trap
Here's where it gets interesting. Bram Van der Hallen (Edge.be, Top 50 PPC Experts Worldwide, who documented this first) writes on LinkedIn:
Editing placements at the ad set level isn't supported after you've selected specific placements at the ad level.
Translation: once you've selected specific placements at ad level, Meta locks ad-set-level editing for that ad set. You can no longer change the ad-set-level selection. Done. If you later realize the whole ad set should run on Reels only, you have to change placements per ad, not centrally.
Practical workflow takeaway:
- Set ad-set-level placements first. Even if you plan to vary at ad level later. If you want to leave "Advantage+ Placements", start with that.
- Only then make ad-level adjustments when you actually need them.
- Not the other way around. Anyone who reflexively clicks at ad level and changes placements there (old UI habit) is locked in.
Not catastrophic when you know. If you run five or ten accounts in parallel like I do, it's a new reflex you need to develop. Especially in hand-over situations with junior buyers or new colleagues, this is a stumbling block.
Update 2: Purchase retention from 180 to 730 days
This is the bigger update.
Before, custom audiences based on Purchase events could capture users up to 180 days back. If you bought in the last six months, you were in. If your purchase was 200 days ago, you fell out.
Now: up to 730 days. Two years. Buyers who haven't done anything in 23 months can theoretically come back into targeting.
What it means in practice
Bigger retargeting audiences. If your product has repeat buyers (supplements, cosmetics, consumables, software subscriptions), this is genuinely valuable. You can track cohorts over two years. "You bought eight months ago, here's the follow-up bundle" becomes feasible without your own CDP.
Better seed audiences for lookalikes. More historical buyer data means better lookalike models. Even though classic lookalikes became less central with Advantage+ Audiences, the longer history helps Meta learn internally. And in the upper funnel that matters.
Bigger exclusion audiences. Past buyers can be excluded from prospecting campaigns for longer. Saves budget at top of funnel where it hurts most. Especially for brands with high CPM pressure, that's real.
Finer segmentation possible. You can now stagger audiences by recency and tune creatives accordingly. My standard setup for a subscription product:
- 0 to 30 days: hot buyers, cross-sell, premium upgrades
- 31 to 90 days: follow-up and onboarding depth
- 91 to 180 days: re-activation, showing new use cases
- 181 to 365 days: quiet win-back, no hard selling
- 366 to 730 days: just remind, don't burn discounts
Five audiences, five creative strategies, five different CPM expectations. Previously you stopped at 181 days. Now it keeps going.
Where it goes wrong
Meta is partially auto-converting existing Purchase Audiences to 730 days. If you do nothing.
Sounds convenient. It's risky.
Concrete scenario. Your audience is called "Buyers last 180 days". Meta silently extends it to 730 days. Suddenly buyers from two years ago are in. You're mixing hot repeat buyers with stale leads.
Your creative strategy no longer matches the audience. CPMs rise because Meta serves broader. Conversion rate drops because many recipients no longer associate your brand with your product. You wonder why a campaign that worked last month suddenly tanks.
The auto-migration is one issue. The second: anyone running a multi-account setup (an agency with 10+ accounts, say) may not have visibility everywhere audiences are being converted. An audit is due now.
What I'd do in your place
First. Go to Ads Manager → Audiences. Check if a banner is showing the auto-conversion notice. If yes, decide per audience whether 730 days makes sense or you want to keep the old window. Leaving everything on 730 blanket is rarely smart.
Second. Audit your existing custom audiences. Did a team member or agency account adjust windows recently without asking? Was a trigger auto-converted? Better to walk through once now than debug performance drops in four weeks.
Third. Use strategically, not blanket. If your setup involves subscription or multi-touchpoint conversion, actually build the recency staggering. If your model is single-sale (wedding venue, insurance signup), 730 days gives you little and costs targeting quality.
Fourth. For consumer goods with short purchase cycles (fashion, decor, fast-moving FMCGs) 180 or 365 days usually stays more useful. Seasonal buyers from 18 months ago have nothing to do with the current collection. Buying back their attention costs more than acquiring new customers.
Who actually benefits
Clear winner. Supplements, beauty, care products, subscription models, anything with a consumption cycle. 730 days enables genuine lifecycle steering through ads.
Often doesn't have direct purchase events in Meta but lead-magnet strategies. Indirectly relevant for longer lead nurturing via audience exclusion. If you do track purchases in B2B: you benefit.
Be careful. Seasonal buyers from 18 months ago have nothing to do with the current collection. 365 days is often the maximum here. Better to opt out of auto-conversion.
Little direct benefit. If a buyer doesn't return (wedding, house, B2B service), longer retention for retargeting is worthless. But: good for exclusion audiences in prospecting.
Interesting for repeat businesses like hair salons, beauty, auto repair. With 730 days you can identify "customer for two years" cleanly without CRM integration.
Two sentences to close
Both updates are small. Both break old setup routines. Anyone just continuing as before will have messy audiences in two weeks and a placement setup that can't be edited cleanly. 30 minutes of audit now saves three hours of debug in a month.
If you're currently reviewing your Meta setup: take the Audiences tab seriously. Walk through every custom audience. Decide consciously what should be 730 days and what shouldn't. This is one of those places where performance marketing discipline pays off immediately.