IVT shows in Beeswax before it shows in the dashboard
After HUMAN FraudSensor was attached, invalid impressions stopped being a permanent zero. Neon rollups still do not store those fields. Viewability is still unmeasurable.
Published 2026-09-10 · Search job: invalid impressions Beeswax not in dashboard
Invalid impressions are a real field on this seat. They stay zero until a measurement vendor is attached. After we attached one, Beeswax started returning non-zero IVT. The dashboard's Neon rollups still do not store those fields.
Until September 8, 2026, every campaign we had run carried creative_addons: []. invalid_impressions and viewable_impressions were 0 because nothing was measuring. We were measuring nothing and reading zero as fine is that note.
We then attached creative add-on 44, HUMAN FraudSensor Banner tag, vendor 7, at $0.05 CPM, on the ten live test creatives. A live performance_agg pull on 2026-09-09 in the Eastern wall clock this seat requests showed:
- Invalid impressions of 0 on every complete day before attachment (September 3 through 7).
- 22 invalid of 5,671 impressions on September 8 (0.388 percent).
- 51 invalid of 4,174 impressions so far on September 9 (1.222 percent).
Anonymized main-seat brand-name display flights we can cite without naming the advertiser sat at 0.523 percent when acceptable and 1.544 percent when they triggered intervention. September 8 is at the acceptable line. September 9 so far sits between acceptable and intervention. That is a window, not a verdict on remnant as a category.
What still does not exist
Viewability is still unmeasurable. Add-on 44 is an IVT tag, not IAS or DoubleVerify. Viewable Impressions stayed 0 and viewability percent stayed null.
IVT is parsed from Beeswax and is not written into impression_log_rollups. A daily script that only reads Neon will still report IVT as missing. Judge current IVT from a live performance_agg pull, not from the wallet snapshot.
Click-to-page-view is still not the deciding metric on this tag-on. Landings and billed clicks can disagree because they are different event definitions. Do not collapse them.
Attach measurement before you optimize remains the plan page for making this a product default. How to judge display traffic is still IVT plus click-to-page-view, never CTR. We will not call a day "low fraud" from a Neon zero.