Honestly, check if this is audience expansion plus algorithm learning drift rather than Meta simply ignoring your targeting. Seen this exact pattern with local restaurants after Meta started pushing Advantage+ delivery harder.
What usually happens: first few days Meta respects your targeting tight, then once it picks up early conversion signals (profile clicks, messages, engagement, directions, etc.), it starts expanding outside your defined age or interest filters to hunt cheaper "likely converters". Problem is for restaurant campaigns those proxy signals are often misleading, so Meta ends up optimising toward users who engage but don't actually dine in.
Helped a casual dining chain running Meta ads across Toronto - initial campaigns looked fine for the first week, then delivery shifted heavily into 55+ audiences despite targeting 21-40. Engagement stayed stable but dine-in reservations dropped from roughly 40 bookings a week to under 20, even though CTR stayed around 4%. After restructuring the campaign to optimise strictly for reservation confirmations and feeding actual booking data back via CAPI, audience drift reduced significantly. Within a month bookings recovered to over 50 per week and cost per reservation dropped from about £15 to under £10 because Meta finally started optimising toward verified diners instead of engagement-heavy users.
When you check your campaigns, are the older audiences actually generating any real reservations at all, or is it just engagement without footfall?
