DSP supply quality in AMC: comparing sites and supply sources

In short: In `dsp_impressions` the fields `site`, `supply_source`, `publisher_name` and `deal_id` show where ads ran; `placement_view_rate` and `supply_source_view_rate` give the viewability rate by MRC standard. That lets you compare inventory by cost and viewability. You then set exclusions in the DSP, not in AMC.

Fields for assessing inventory

  • site: website, app or streaming platform; app_bundle the app identifier; publisher_name the publisher.
  • supply_source: where the impression was bought (Amazon properties, Amazon Publisher Services, partners).
  • deal_id and deal_name: for purchases through Private Marketplace or Programmatic Guaranteed; NULL for open auction.
  • placement_view_rate and supply_source_view_rate: viewability rate between 0.0 and 1.0 by MRC standard (at least 50 percent of pixels for at least 1 second for display, 2 seconds for video). placement_is_view_aware and supply_source_is_view_aware show whether the source measures viewability at all.
  • total_cost in millicents (divide by 100,000) includes all fees; supply_cost in microcents (divide by 100,000,000) is the price paid to the publisher.

Query: cost and viewability per site

SELECT
  site,
  supply_source,
  SUM(impressions) AS impressions,
  SUM(total_cost) / 100000 AS cost_in_currency,
  AVG(supply_source_view_rate) AS avg_view_rate
FROM
  dsp_impressions
GROUP BY
  site,
  supply_source
The example has not been run on an AMC instance. For conversions per site you must join the attribution tables via the event identifier (see the schema reference).

Our take

Few impressions per site give unreliable rates, and small groups drop out through the aggregation thresholds. We therefore assess inventory by groups of sites rather than single sites. That is our recommendation, not Amazon advice.

Sources: Amazon Ads API: Amazon DSP impressions (retrieved 2026-10-09)

Keep reading: Ad type Amazon DSP · Schema reference