BenchmarksArticleSeat-measured

Supply Bench wave scorecard

Dated remnant scorecard from August 2026 Supply Bench waves. Partial coverage. No IVT vendor. Not an industry average.

Published 2026-09-09 · Search job: remnant display placement waste

Last measured: 2026-09-02, read-only inventory_agg and domain_agg for Supply Bench waves flown in August 2026. This is a seat-measured slice, not a category benchmark. If a later covered pull disagrees, we will edit this page and date the change.

Coverage (read this first)

This seat ignores report offset. A 50,000-row limit returns 400. The pull kept the top rows by spend: about $316.64 of a roughly $432 combined-wave total on the placement view, and $330.30 on the domain view. Every share below is biased toward larger placements. The missing tail is many tiny rows. Partial coverage is part of the finding.

Campaign-grain dashboards could not answer placement, domain, seller, or SSP identity. We pulled those views because the exchange had already billed them.

No IVT vendor was attached. invalid_impressions and viewable_impressions are not a quality score on this flight. Zero means nothing was measuring.

Placement and domain grain

SignalNumberReading
Placements in the slice1,149Long tail, not a short list of slots
Domains in the slice1,535Same spray at domain grain
Zero-click placements93.4 percent (1,073)Most slots did nothing measurable
Spend on those zero-click placements$202.77, 64 percent of observed spendThe money followed the empty slots
Click concentration7 domains = 41.3 percent of clicks from 4.7 percent of impressionsBilled clicks are concentrated, not evenly earned
Unauthorized ads.txt spend$0.48ads.txt would not have saved this buy
Reseller vs direct CTR0.423 percent vs 0.184 percentResold paths clicked 2.3 times more
SSP CTR range (same creative)OPENX 0.111 percent to SMARTADSERVER 3.074 percentSupply behavior, not a new banner

Do not quote the billed click total as engagement. Roughly two fifths of clicks in this slice sit on the seven-domain set.

What we will not claim

We will not say the flight was high fraud or low fraud. We will not treat 93.4 percent zero-click placements as a universal remnant constant. We will not promote ads.txt as the fix. We will not turn OPENX's 0.111 percent into "never buy that exchange" without a later judged exclude.

environments=['web'] held (app_agg returned 0 rows). Browser games and utilities still appeared. Web-only is not "no casual inventory."

How to use this scorecard

Compare a new covered pull to these rows. If coverage is again partial, say so. Judge the next flight on concentration and click-to-page-view, and on IVT only after a vendor tag is on.

The watchlist of repeat domains is click contamination watchlist. How we read any remnant report is how to judge display traffic.