How manual placements work
An ad set can either let Meta spend across every available surface or be restricted to the ones you tick. Choosing the second is manual placements: you open the list of placements and select the feeds, stories, reels, Marketplace, in-stream video, Messenger or Audience Network positions you are willing to buy, and delivery is confined to those.
The list is not the same for every campaign. Objectives, formats and the app you are advertising all affect which surfaces are eligible, so a placement you used on one campaign may simply not appear on another. Nothing about the choice is permanent — it is an ad set setting you can loosen at any time, and loosening it later is usually easier than starting narrow and trying to explain poor delivery.
Why it matters
Every surface you switch off removes inventory from the auction the delivery system can buy. A narrower auction means fewer chances to find a cheap impression for the right person, so the usual outcome of heavy restriction is a higher cost per result and a slower, less stable learning phase. That is the trade you are making, and it should be a deliberate one.
There are still good reasons to make it. Brand safety concerns about third-party apps, a creative that only exists in one shape, a legal or client requirement to stay inside a named set of surfaces, or a placement that has repeatedly produced enquiries your sales team cannot use — all of those justify taking the choice back.
Where manual placements go wrong
The commonest mistake is restricting the ad set to solve a creative problem. If the story version looks wrong, the fix is a story-shaped asset, not switching stories off — and placement asset customisation lets you keep every surface while controlling exactly what appears on each one.
The second is judging a surface too early, on impression cost rather than on results, and cutting it before it has produced enough data to mean anything. The third is forgetting the setting exists. A restriction added for a campaign months ago quietly follows every ad set duplicated from it, and nobody notices until someone asks why reach has stopped growing.
How to act on it
Default to letting delivery use everything, and let the placement breakdown in reporting tell you where your ads actually work. If a surface looks poor, check the creative fitted it properly before you conclude anything, and wait for a meaningful number of results rather than a couple of days of spend.
When you do restrict, restrict for a written reason and review it on a date you set. Keep the restriction as small as it can be — removing one problem surface rather than ticking only your favourites — and supply properly shaped ad creative for everything you keep. If cost per result rises after a restriction, that is the narrower auction talking, and widening it again is the first thing to try.
