eCPM ranges, with the method attached.
An index of eCPM ranges by format, geography and platform, for in-app and CTV inventory. Published quarterly. Methodology and caveats are part of the report, not a footnote, because a benchmark without its sample is a rumor.
A benchmark with its sample attached
Ranges, not point estimates. Definitions included.
By format
Rewarded, interstitial, banner and native in-app; pre-roll and mid-roll on CTV. Each with the impression definition used for the range.
By geography
Tiered regions rather than single-country claims, with the tier definitions published so the ranges can be applied to your own mix.
By platform
iOS versus Android in-app, and per TV platform group on CTV, stated separately rather than blended into one misleading average.
Methodology
Source
Aggregated network data from Sky Boosts inventory where the sample is sufficient, supplemented with clearly labeled third-party sources where it is not. The report states which is which, per row.
Period
Each edition covers one quarter and states its window. No edition silently reuses an older period.
Definitions
Every figure carries its impression definition and currency. A range without a definition is excluded from the report rather than normalized.
Thin samples
Where first-party data is thin for a row, the report says so and either labels the third-party source or omits the row. Thin data is disclosed, not dressed.
Latest edition and archive
The first edition is in preparation. Access is by email: the report is sent when published, and each later edition arrives as it is released. The email is used for the report and its updates only.
Archive: editions are listed here by period once the first exists. Nothing is listed yet, and the list stays honest rather than padded.
Citation policy
Figures may be reused with attribution to Sky Boosts and a link to this page, including the edition period. Charts may be reproduced unaltered. Quoting a range without its definition is not a permitted citation.
Benchmarking, evaluating, negotiating.
Benchmark a waterfall
Compare placement-level eCPM against the format and geo ranges, and let the outliers direct attention instead of averages.
Evaluate a mediation switch
Use the ranges as the like-for-like baseline for the parallel test, so the comparison does not inherit the old setup's ceiling.
Negotiate floors
Floor proposals are strongest when they cite a range and its definition. The report gives you both, per format and tier.
Caveats: sample bias exists and is described per edition; format definitions differ across networks and are stated here explicitly; your geo mix can move your real numbers outside any published range. The report is a compass, not a contract.
Get the report when it publishes.
Email address only, used for the report and its edition updates.
Send a routed message
ROUTE: PUBLISHERBenchmarks in practice.
Yes. Access to each edition is by email so the report goes to people who want it, and the list is used for nothing else.
No. The report publishes ranges with definitions and sample sizes per row. Raw impression-level data is not distributed.
Attribute Sky Boosts, link to this page, and include the edition period. Quoting a range without its definition is not a permitted citation.
No. Rows only publish where the sample is sufficient for aggregation, and no publisher-level data appears in any edition.