Routing
Publifter doesn't touch your mediation setup. You keep your own ad units and floors; Publifter decides which of your existing units to request for each user, based on their value tier.
Example: your mediation already has a high-floor waterfall and a low-floor waterfall. A Premium user's request goes to the high-floor unit; a Dormant user's goes to the open unit. Nothing about the units themselves changes.
71% of active users are on an app version carrying the SDK. Per-user figures cover those users only.
Floor simulator
Set an eCPM floor per tier. A floor above what that tier's impressions clear at raises the average price and costs fill. Stated model, not a live yield curve.
Targets Ad Unit — High Floor (Waterfall A).
Targets Ad Unit — Standard Floor (Waterfall B).
Targets Ad Unit — Low Floor (Waterfall C).
Targets Ad Unit — Open (No Floor, Waterfall D).
Model: for every 1% a floor sits above that tier's clearing price, the average price paid rises 60% as fast and fill falls with the square of the gap, floored at 35% fill. Baseline is your current single floor of $0.42 applied to every tier.
Modelled, not measured. See modelled vs actual below.
At today's single floor.
At the floors set above.
Share of traffic covered by active rules below.
Per-tier detail
Recomputed live from the sliders and active rules.
| Tier | Floor | Projected fill | Revenue / user | Baseline revenue / user | Monthly lift |
|---|---|---|---|---|---|
| Premium | $2.30 | 91.8% | $0.139 | $0.128 | +8.5% |
| Core | $0.80 | 87.8% | $0.044 | $0.041 | +8.5% |
| Casual | $0.34 | 90.1% | $0.012 | $0.011 | +7.4% |
| Dormant | $0.13 | 89.3% | $0.002 | $0.002 | +5.1% |
Projected revenue per user by tier
Solid bars are the single-floor baseline you run today; the second bar is projected at your simulated floors.
Routing rules
Each rule targets one of your existing mediation ad units. Toggle a rule off and its traffic falls back to the baseline floor in the totals above.
| Rule | Condition | Target ad unit | Floor | Traffic share | Status | Last changed |
|---|---|---|---|---|---|---|
| Premium — US rewarded | Premium · United States · Rewarded video | Ad Unit — High Floor (Waterfall A) | $2.30 | 3.1% | Active | Sep 10, 2026 |
| Premium — ROW rewarded | Premium · Rest of world · Rewarded video | Ad Unit — High Floor (Waterfall A) | $2.30 | 2.9% | Active | Sep 10, 2026 |
| Core — interstitial, all geo | Core · All · Interstitial | Ad Unit — Standard Floor (Waterfall B) | $0.80 | 18.4% | Active | Sep 6, 2026 |
| Core — banner, all geo | Core · All · Banner | Ad Unit — Standard Floor (Waterfall B) | $0.80 | 12.6% | Active | Sep 6, 2026 |
| Casual — banner, all geo | Casual · All · Banner | Ad Unit — Low Floor (Waterfall C) | $0.34 | 27.8% | Active | Aug 29, 2026 |
| Casual — native, all geo | Casual · All · Native | Ad Unit — Low Floor (Waterfall C) | $0.34 | 13.2% | Active | Aug 29, 2026 |
| Dormant — open waterfall | Dormant · All · All formats | Ad Unit — Open (No Floor, Waterfall D) | $0.13 | 22.0% | Active | Aug 20, 2026 |
Modelled vs actual lift
Modelled lift comes from the elasticity above. Actual lift is measured revenue over the trailing 4 weeks versus the pre-Publifter single-floor baseline.
| Tier | Modelled lift | Actual lift (4wk) | Variance | Why they differ |
|---|---|---|---|---|
| Premium | +24.5% | +21.2% | -3.3pp | Actual trails modelled: some premium users churn demand at $1.60 floor in thin geos, lowering realized fill versus the elasticity assumption. |
| Core | +11.0% | +12.4% | +1.4pp | Actual beats modelled: bidder density in core geos was higher than assumed, so fill held up better than the flat elasticity predicts. |
| Casual | +6.5% | +5.1% | -1.4pp | Actual trails modelled: casual users skew toward lower-CPM regions where demand is thinner than the blended elasticity assumes. |
| Dormant | +1.8% | +2.0% | +0.2pp | Close to modelled: at a near-zero floor there is little elasticity to misjudge. |
Impressions on demand without user-level revenue (Meta Audience Network, AdMob Partner Bidding, AdSense/AdX web) are still requested per the fallback unit for a user's tier, but their revenue is excluded from the modelled figures above and reconciled separately at the cohort level.