BenchmarksPlanPlan

Grain-level waste mix

Recurring plan. After each covered Supply Audit run, publish spend mix by exchange, placement, and site. This page does not invent a new run.

Published 2026-09-09 · Search job: programmatic placement exclusions

This page is a plan, not a result. We have not judged this test yet. The protocol is here so the next flight can be read. Numbers below are setup notes or prior adjacent evidence, never a claim that this test already ran.

Status: plan. Republish after each Supply Audit run whose coverage_verdict is covered. Do not publish a mix from a no_account_rows run or from a local script that could not see the tenant.

This page is the slot for grain-level waste after the product scan, not a second copy of the August 2026 Supply Bench scorecard. Those waves were a reporting pull, not an audit-apply cycle. Their numbers stay on Supply Bench wave scorecard. We will not reuse 93.4 percent, $202.77, or 41.3 percent here as if they were the latest audit mix.

What a covered run must include

  • Window and account-level spend that the run actually observed, using per-probe Math.max so inventory_agg and domain_agg are not summed as two budgets.
  • Coverage verdict and the row-cap or offset limit, in plain language.
  • Mix by inventory source, placement, and site. Those are the grains the apply path can act on. Domain remains a list-object grain and must be labeled blocked if the run could not apply it.
  • Whether each finding is open, applied, or dismissed. Applied does not mean live delivery changed. Apply creates targeting. It does not re-point the campaign.
  • IVT left blank unless a vendor tag is on and quality fields are non-zero.

What we will not do

We will not invent a 2026-09-09 audit mix to fill this template. We will not turn a partial pull into a pie chart that looks complete. We will not name advertisers.

When the next covered run exists, replace the plan body with a dated table and keep this methods section above it.