Meta Audience Network placements: exclude them for lead gen, or keep them for volume?

Audience Network and the extra Facebook feed placements tend to show cheap clicks, and cheap clicks are easy to mistake for good traffic. For lead gen with a clean form event, I'd exclude Audience Network first and compare cost per lead over two weeks. The usual result is clicks that never fill the form.

I'm not sure that holds for e-commerce with a small catalog, though, where those placements can be what lets the budget actually spend. A dental clinic running $30 a day should cut the network placement. A store with a few hundred SKUs struggling to exit learning should probably keep it and judge by purchases, not CTR. Which kind of account are you running?

Agreed on the split, but for lead gen there's one more thing to check. Audience Network leads often come from accidental taps inside apps, so the forms fill up and CPL looks great, while the people behind those fills never really meant to enquire. Pull the placement breakdown and compare cost per qualified lead, or at least lead-to-booking rate, instead of CPL. If the network leads never book, excluding it is the right call. If they book at a similar rate to feed leads, the cheap CPL was real and you keep it. Either way, the decision should come from the booking data, not the click cost.

When you split by placement, do the cheap leads ever turn into booked appointments?

Ads Manager will split leads by placement, but it has no idea which ones ever booked. To fix that, store the form's lead ID in your CRM so you can match bookings back to the placement that produced each lead.

Give the booking lag some room too. If a dental enquiry often books nine days after the form fill, a two-week read will make the network leads look dead when they just haven't converted yet. And once booked appointments are sent back through the Conversions API, Meta can optimize toward those instead of raw fills.

The lead ID only helps if it lands in the CRM automatically. If the front desk types leads in by hand, the ID gets lost and you're back to matching on name and phone, which is a mess with dental enquiries since half the callers book under a spouse's number. Use the native CRM integration or a connector so the leadgen ID is saved on the record at submit.

On the Conversions API side, the booking event needs that same lead ID in user_data, or Meta can't tie the appointment back to the original Audience Network fill. Send it as an offline event stamped with the booking date, not the form date. Otherwise your lag analysis and Meta's optimization will disagree about when each lead converted, and the network leads will look better or worse depending on which clock you trust.

It comes down to how many bookings the account produces each week. Meta needs a steady flow of conversion events per ad set to learn from, and the usual rule of thumb is around 50 a week. A dental clinic booking eight or ten new patients a month from Audience Network won't get anywhere near that, so sending offline booking events mostly helps your spreadsheet, not Meta's delivery. In that case, keep the lead ID matching for your own lag analysis and let Meta keep optimizing on the form event. Once booked volume gets close to that weekly number, the offline signal starts to affect who sees the ads. So pull the weekly booking count first, then decide how much effort the Conversions API setup deserves.

The 50 a week threshold applies per ad set, not per account. A clinic with three ad sets splitting $30 a day is much further from it than the account total suggests. It's worth pulling the offline booking count for each ad set in Events Manager over the last seven days, since that's the data Meta actually learns from, and it can differ from what your CRM shows. Keep in mind that Meta reads the booking date, so a Friday booking for a Tuesday fill counts on Friday. If each ad set only gets a trickle of booking events, merging them into fewer ad sets will probably do more for delivery than a more elaborate Conversions API setup.