Meta Audience Network in Advantage+ placements: switch it off for lead gen?

Picture a dental clinic running lead gen on a modest budget with Advantage+ placements left on automatic. The reports show cheap clicks coming from Audience Network, the cost per lead looks great, and yet those form fills rarely turn into booked appointments. My take is that Audience Network is mostly a reach dump for a service business like this. A lot of those placements are accidental taps in games and apps, so the people filling out forms have much lower intent than someone scrolling Feed or Reels.

I'd exclude Audience Network, keep the Instagram and Facebook feeds, and judge results by cost per booked appointment over a few weeks instead of cost per lead. The catch is volume. If the budget is small, cutting a placement can slow delivery and you won't learn much from the test. Has anyone run a clean split on this with real lead quality data?

Hold off on blaming Audience Network until you split it out. Prefilled instant forms get accidental fills in any placement, so the gap could be coming from the form rather than the placement. Import booked appointments as offline conversions and compare booked rate by placement before you cut anything.

I'm not sure the form explains it as neatly as you're suggesting. Prefilled instant forms do get accidental submits, but the same form runs in Feed, Reels and Audience Network. So if the gap shows up in one placement and not the others, the form is the constant and the placement is the variable.

Where I agree is that accidental taps aren't evenly spread. Audience Network has a lot of game and app inventory where a thumb lands on an ad by mistake, so the accidental-tap rate there is probably much higher than in Feed, which is basically Mira's point. Splitting by placement with offline conversions is still the right test. Just don't expect it to clear the form if the booked rate drops only in Audience Network.

The form stays constant, so the placement split is still the test worth running. Once you have offline conversions, stop judging placements on cost per lead and judge them on cost per booked appointment. Cheap leads from a placement can look like a win right up until you see the booked rate, and that's where the real difference shows up.

Sample size is the other half. For a small clinic, a placement can look terrible after 8 booked appointments and fine after 25, so I'd wait for roughly 20 to 30 booked appointments on each side before cutting anything. If the budget won't get you there in a reasonable time, don't exclude Audience Network outright. Put it in its own ad set with a capped budget. That keeps the test clean without starving delivery in the placements you actually want, and you still get a readable answer.

Waiting for 20 to 30 booked appointments per side doesn't make sense for a clinic on a modest budget. That could mean months of paying for a placement you already suspect is weak, and the call is easy to undo, so you don't need full certainty before acting. Pick the most you'd pay per booked appointment, and if Audience Network goes past that after a fixed spend, exclude it. The 48-hour contact rate on form fills is a good early signal too, because that gap usually shows up well before bookings do.

Agreed on having a cutoff, but I'd base it on cost per booked appointment, and only once the lag has passed. Form fills from a cold placement often get called back days later, so a fixed spend window can kill a placement right before its bookings land. Work out the window from how long your past inbound leads took to book, then wait until most of that cohort has had time to show up.

I'd also double-check whether a placement that books at the same rate but costs more per lead is really a loser. It might just be the expensive one.

When a booking happens, does the lead's ad source survive into your CRM, or are you matching by phone number?