Skip to content
T1KTheOneKit

t1k:marketing:monetization:mediation-strategy

FieldValue
Modulemonetization
Version1.16.9
Efforthigh
Tools—

How to invoke

Section titled “How to invoke”
/t1k:marketing:monetization:mediation-strategy

Mediation Strategy

Section titled “Mediation Strategy”

Cross-network mediation decisions and the per-network reporting unification pattern that makes them measurable.

When to Use

Section titled “When to Use”
  • Choosing/comparing a mediation platform (AdMob vs AppLovin MAX vs Unity LevelPlay)
  • Deciding waterfall vs in-app bidding (IAB) and the network mix
  • Setting eCPM floors, ARPDAU targets, and reconciling revenue across networks

Mediation platform comparison

Section titled “Mediation platform comparison”
PlatformPer-network reporting dimensionBiddingNotes
AppLovin MAXnetwork (⊥ requests → two-call)strong IABestimated_revenue = whole-mediation total. See [applovin-max].
Google AdMobAD_SOURCE / AD_SOURCE_INSTANCEbidding + waterfallearnings in micros; no requests/network conflict. See [admob].
Unity LevelPlay (ex-ironSource)adSourceIAB20 req / 10 min reporting limit. See [ironsource].

Waterfall vs in-app bidding

Section titled “Waterfall vs in-app bidding”
  • Waterfall: networks ranked by historical eCPM; called in order until filled. Simple, but stale floors leave money on the table.
  • In-app bidding (IAB): all bidders compete in a unified real-time auction. Higher yield, less manual floor tuning; prefer IAB-eligible networks (MAX, AdMob, FAN, LevelPlay) and keep a thin waterfall only for non-bidding sources.
  • Hybrid is normal: bidding auction + a short waterfall tail.

The reporting-unification rule (critical for dashboards)

Section titled “The reporting-unification rule (critical for dashboards)”

Every mediation platform reports a whole-mediation total plus a per-network decomposition under a vendor-specific dimension (MAX network, AdMob AD_SOURCE, LevelPlay adSource, FAN as one source). To compare networks apples-to-apples:

  1. Pull each platform’s per-network report and normalise the network dimension to a single network field in your warehouse.
  2. Enforce the invariant SUM(per-network revenue) == platform whole-mediation total per (day, app) — a per-network sum that drifts from the headline total signals a parse/merge bug (e.g. the AppLovin merge regression that wrote network='unknown'; see plan 260602-2108-adnet-per-network-revenue).
  3. Watch for dimension conflicts (MAX network ⊥ requests) that force multi-call designs — never multiply request counts across networks.

eCPM / ARPDAU optimisation

Section titled “eCPM / ARPDAU optimisation”
  • Track ARPDAU = ad_revenue / DAU per network and placement; raise floors where fill is high and eCPM is rising, lower where fill starves.
  • A/B test waterfall vs bidding and floor levels via each platform’s experiment tooling; decide on ARPDAU + retention, not eCPM alone (high floors can cut fill and net revenue).
  • First-look / bid-floor anchoring: set a first-look auction floor at the P90 CPM (impression-weighted for aggregated mediation data — never a naive percentile(ecpm, 90)), not the average. See iaa:mediation-waterfall-strategy → first-look-p90-floor.

References

Section titled “References”