Price-floor diagnosis

Revenue dropped after changing price floors

A higher floor can raise the price of some winning impressions while reducing eligible demand or total fills. Diagnose total revenue per request and per impression before deciding whether a floor change helped.

Do not judge only by eCPM

An eCPM lift can still produce less total revenue if requests, fills, or impressions fall more sharply. Compare revenue, weighted eCPM, fills, and impressions together.

Check the affected rollout segments

Compare countries, ad units, formats, and sources that received the floor change against unchanged or comparable segments. Broad averages hide where the trade-off occurred.

Use comparable windows

A single day can be distorted by weekday patterns, seasonality, demand budgets, or traffic mix. Use matching periods and record the exact floor and rollout time.

  1. Record the previous floor, new floor, rollout time, and affected segments.
  2. Compare requests, fills, impressions, weighted eCPM, and revenue per 1,000 requests before and after.
  3. Split results by country, format, placement or ad unit, and ad source.
  4. Check whether the eCPM lift was offset by lower fills or impressions.
  5. Compare changed segments with a controlled or unchanged segment where possible.
  6. Avoid repeatedly moving floors before the trade-off is visible in the report.

Common interpretation

If weighted eCPM rose but total revenue fell, the floor may have reduced the volume of eligible or filled opportunities. If both eCPM and fill fell, source demand, traffic mix, or a broader configuration change may need separate investigation.

Use a matching before/after CSV to decide whether the floor changed pricing, fill, traffic composition, or several signals at once.

Start with the sample diagnosis, then escalate when needed.

Use anonymized before/after data. Do not send account access, API keys, or private identifiers.

Try demo with sample dataRequest free diagnosis