given the chaos of the past week, I'm glad to see some stabilisation on our end too. we manage about 15 ad accounts across different verticals, and after a brutal Monday-Wednesday window, delivery metrics finally converged back to baseline around Thursday evening.
A few things I've been tracking through our custom dashboard (pulling from the Ads API + a lightweight Python script for hourly snapshots):
- Impressions served - back within 5% of the pre‑outage average. The "delivery by daypart" graph stopped looking like a sawtooth wave.
- CPA variance - shrank from ±40% to ±8% over a 48‑hour rolling window. Still a little jittery on some lower‑spend campaigns, but manageable.
- Frequency - normalised across the board. No more one‑user‑gets‑twenty‑ads nonsense.
I'd still keep a close eye on attribution windows for the next few days - some of our clients' conversions took a hit on Tuesday, and the delayed reporting meant we were flying blind through optimisation. If you haven't already, worth cross‑referencing with Google Analytics or your own server‑side events to make sure the platform's data is aligning again.
The bloodbath seems over, but I'm not turning off the monitoring script until we hit a full week of stable data.
