LabArticleOperator-judged

Open exchange waste, mapped by grain

There is no Beeswax object named supply route. Waste shows up in reports by placement, site, domain, and exchange, and only some of those grains can be excluded today.

Published 2026-09-09 · Search job: open exchange vs private marketplace

Open exchange versus private marketplace is the wrong first split if you cannot name the grain you will act on. There is no Beeswax object named supply route. What people call a route is targeting plus lists plus deals plus modifiers plus reports, assembled after the fact. On the open exchange, waste is something you read back by placement, site, domain, and inventory source. A private marketplace is a deal object this child seat has historically not used.

This page is the map. The measured waste sits in the companion articles. The deal comparison is a plan, not a result.

Grains the report can show

The 2026-09-02 Supply Bench pull of inventory_agg and domain_agg for campaigns 4791 through 4810 is the measured slice we will keep citing:

  • 1,149 placements, 1,535 domains.
  • 1,073 placements (93.4 percent) had zero clicks and took $202.77, 64 percent of observed spend.
  • Top ten domains took 31.9 percent of spend.
  • Seven domains produced 41.3 percent of clicks from 4.7 percent of impressions.
  • OPENX took 44 percent of spend at 0.111 percent CTR.
  • ads.txt Unauthorized was $0.48 against $195.56 Direct and $120.59 Reseller.

Coverage was the top roughly 5,000 rows by spend, $316.64 of a combined total near $432. Offset is ignored on this seat. A 50,000-row limit returns 400. Filters on campaign_id fail.

That is open-exchange waste at four grains: placement, domain, inventory source, and ads.txt authorization. Site appears as a separate report column and as a targeting key. Private marketplace waste is not on this map because we have no deal delivery to read.

Grains the seat can exclude

Placement and site are direct targeting keys (related_field is key, comparator equals). Domain and app_bundle need seat-level reference lists. inventory_source is an allowlist. Empty compiles to match nothing. The old sentence that placement and site also needed a custom_list was wrong and it blocked 59 percent of findable waste on the first real Supply Audit run: $47.71 placement plus $34.21 site of $138.53, versus $7.79 the product could act on.

The how-to is Placement vs site vs domain. The allowlist trap is Why an empty allowlist serves nothing. Domain after an exchange filter is still a list-object problem: Domain waste after we filter exchanges.

How to read this against a PMP

A private marketplace is a deal you attach, not a cleaner open exchange. We will not write that deals would have removed the 1,073 zero-click placements. We have not flown that test. We will write that open exchange is where this seat has all of its measured delivery, that waste is concentrated by domain and placement, and that the useful question after a scan is which grain you can actually compile.

QPS is not on this map. Front-end QPS filters decide which auctions enter the bidder. Line-item targeting decides which already-received requests may bid. Excluding a placement in targeting does not free QPS. That distinction is QPS is not targeting.

For the vocabulary, see What we mean by a supply path. For the placement pile, see Most placements we bought did nothing.