advantage+ is the default, IPM signal is noisier, and the old bidding heuristics overshoot. our read from live accounts — and what to change monday.
meta rolled advantage+ shopping and app campaigns to default through Q2, and the downstream effects are now clear enough to act on. this is our read from live production accounts — what changed, what broke, and what to actually do about it.
what changed
the headline is control. advantage+ now decides more of placement, audience, and delivery than it did a year ago, and the manual levers studios used to pull have fewer notches. the trade is real: broader automated reach in exchange for less granular, noisier ad-set-level signal. for UA teams that built their whole process on hand-tuning, that's a bigger adjustment than the release notes admit.
what we're seeing in the data
post-change CPI variance is up about 18% across our accounts. IPM reads noisier day to day, which makes early creative calls riskier — a variant can look dead on day one and recover by day three. and the old bidding rules — cap CPI at target × 1.15, walk it down 10% a week — now overshoot in both directions, because they were tuned for a calmer signal than the one we have.
how we adjusted
we stopped treating variance as noise and started treating it as a first-class signal. the bid controller uses a wider band and a slower decay, so it doesn't yank budget on a single bad read. creative decisions wait for a three-day window instead of a one-day flinch. on the accounts we've re-tuned this way, delivery is back within about ±4% of target.
what to change monday
- widen your CPI guardrails — the old ×1.15 cap is too tight for the new variance.
- give new creative three days before you judge it; day-one IPM is lying more than it used to.
- lean harder on creative volume and grounding, since that's the lever advantage+ still hands you.
numu already reads variance as signal and holds the wider band automatically, so you're not re-learning meta's mood every quarter by hand. see it on your account.
