Advantage+ placements on a small Meta budget: let Meta choose or pick manually?

Say you're testing a new offer on about $20 a day in one ad set. Advantage+ placements looks like the obvious default, since Meta can push spend toward whatever is converting. But at that budget the algorithm can dump almost everything into Reels or Stories, and then you never find out whether Feed would have done better for that offer.

I'd start with manual placements for the first couple of weeks, Feed plus Reels, and only open it up once you can see which placement actually gets cheap results. You give up a little efficiency early, but you get a test you can actually read. Has anyone seen Advantage+ placements clearly help under roughly $50 a day, or does it mostly hide the answer?

Sasha, I'd still leave Advantage+ placements on at $20 a day. Say a dental clinic pays around $25 per booked call. At that budget it might get three or four bookings a week in total. Split that across Feed and Reels and each placement sees one or two conversions in two weeks, which is mostly noise. Each split ad set also gets fewer optimization events, so it can sit in learning longer than the test runs. You end up with a cleaner setup and a less useful answer.

I'd let it run and check the placement breakdown once there are a few weeks of conversions behind it. If one placement is clearly costing more per booking, exclude it then. If you need an answer sooner, test two creatives for the same offer instead. That gives you a clearer read than splitting the same small budget across placements.

Marco, the split doesn't have to mean separate ad sets. One ad set restricted to Feed and Reels still gets a single learning phase, and the placement breakdown in Ads Manager shows where the bookings came from. It's only when you split into multiple ad sets that the learning gets fragmented.

Even with one ad set and manual placements, $20 a day doesn't give you enough data. The placement breakdown only shows where the bookings landed, and with three or four a week that's mostly noise. Feed might show two bookings and Reels one, which tells you nothing about which placement is actually cheaper.

Mira, you're right that the breakdown is noisy at $20 a day, but it mostly comes down to how many bookings you get per week. At three or four, Feed versus Reels won't be readable. At ten or more, which can happen with a strong offer and a well-matched audience, the split starts to mean something within two or three weeks.

The size of the gap matters too. If one placement costs twice as much per booking, you'll see it quickly even at low volume. If the difference is around 10 percent, no amount of data this budget can buy will show it, so the placement barely matters and you can just go with whatever's cheaper to run.

So count your bookings per week and compare that to what you'd need to read a result. That number tells you whether the placement breakdown is worth looking at.

Sasha, I don't think the "twice as much" test holds up. At three or four bookings a week, a 2x gap in cost per booking can easily be random, since one lucky or unlucky booking moves the number by 30 percent or more. It looks obvious on a dashboard, but it tells you very little about which placement is actually cheaper.

Agreed, and I'd push it further. At three or four bookings a week, even the placement totals swing too much to call a winner from one snapshot. What matters is whether the gap holds over time. If Reels is cheaper in week one, Feed in week two, and Reels again in week three, that's noise. If one placement stays worse for three or four weeks running, that's a pattern you can act on. A simple weekly sheet with cost per booking by placement is enough to tell which one you're looking at.