Rebill survival halves by month 3
Rebill survival = share of subscribers still billing at month N. Trial-to-subscription offers, n=14 campaigns.
— M1: 100% ▇▇▇▇▇▇▇▇▇▇
— M2: 71% ▇▇▇▇▇▇▇
— M3: 48% ▇▇▇▇▇
— M4: 34% ▇▇▇
— M6: 19% ▇▇
Networks that pay full rebill commission shift chargeback risk onto you; networks paying CPA-only on the front end do not. The curve decides which model wins.
Read: only chase rebill commission when the M3 survival clears ~45%.
n=14 campaigns
Rebill survival = share of subscribers still billing at month N. Trial-to-subscription offers, n=14 campaigns.
— M1: 100% ▇▇▇▇▇▇▇▇▇▇
— M2: 71% ▇▇▇▇▇▇▇
— M3: 48% ▇▇▇▇▇
— M4: 34% ▇▇▇
— M6: 19% ▇▇
Networks that pay full rebill commission shift chargeback risk onto you; networks paying CPA-only on the front end do not. The curve decides which model wins.
Read: only chase rebill commission when the M3 survival clears ~45%.
n=14 campaigns
Hidden scrub shows up as a stable ~8% conversion gap
Scrub = silently withheld conversions. Detected by comparing your tracker postbacks to network-reported conversions.
— Honest variance from attribution lag: ±2–3%
— Persistent one-directional gap: scrub signal
— Sampled flag threshold: gap >5% sustained 14+ days
In the sample, 6 of 31 networks showed a stable gap above 5% (n=31). Random attribution noise reverses direction day to day; scrub does not.
Read: a one-directional gap that never flips toward your numbers is the tell.
Scrub = silently withheld conversions. Detected by comparing your tracker postbacks to network-reported conversions.
— Honest variance from attribution lag: ±2–3%
— Persistent one-directional gap: scrub signal
— Sampled flag threshold: gap >5% sustained 14+ days
In the sample, 6 of 31 networks showed a stable gap above 5% (n=31). Random attribution noise reverses direction day to day; scrub does not.
Read: a one-directional gap that never flips toward your numbers is the tell.
Effective CPM ranges 9x across verticals
eCPM = revenue per 1,000 impressions, back-calculated from EPC × CTR, 28-day window.
— iGaming: $18.40 ▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇
— Finance: $11.20 ▇▇▇▇▇▇▇▇▇▇▇
— Dating: $6.80 ▇▇▇▇▇▇▇
— Nutra: $4.10 ▇▇▇▇
— Sweeps: $2.00 ▇▇
iGaming's lead is driven by high payout-per-action, not click volume; nutra makes it up on conversion rate and rebill depth (n=19 traffic sources).
Read: a low-eCPM vertical can still win on volume and back-end rebills.
n=19 sources
eCPM = revenue per 1,000 impressions, back-calculated from EPC × CTR, 28-day window.
— iGaming: $18.40 ▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇
— Finance: $11.20 ▇▇▇▇▇▇▇▇▇▇▇
— Dating: $6.80 ▇▇▇▇▇▇▇
— Nutra: $4.10 ▇▇▇▇
— Sweeps: $2.00 ▇▇
iGaming's lead is driven by high payout-per-action, not click volume; nutra makes it up on conversion rate and rebill depth (n=19 traffic sources).
Read: a low-eCPM vertical can still win on volume and back-end rebills.
n=19 sources
Set up a chargeback-rate tracker that survives delays
Chargeback rate = reversed conversions ÷ approved, but timing breaks naive counts.
— Cohort conversions by week, never mix vintages
— Wait the full clawback window before declaring a cohort's final rate
— Compute rate at 30/60/90 days to see the curve flatten
— Rebill verticals: expect the 90-day rate ~1.6x the 30-day read
— Flag any cohort breaching 8% — investigate source quality first
Chargeback maturation (rebill cohort, n=2,100):
30d: 4.1% ▏▏ · 90d: 6.7% ▏▏▏▏
Read: a 30-day chargeback rate is a down-payment on the truth, not the truth.
—
Ещё больше cpm benchmarks в @pixelprofit_fb
Chargeback rate = reversed conversions ÷ approved, but timing breaks naive counts.
— Cohort conversions by week, never mix vintages
— Wait the full clawback window before declaring a cohort's final rate
— Compute rate at 30/60/90 days to see the curve flatten
— Rebill verticals: expect the 90-day rate ~1.6x the 30-day read
— Flag any cohort breaching 8% — investigate source quality first
Chargeback maturation (rebill cohort, n=2,100):
30d: 4.1% ▏▏ · 90d: 6.7% ▏▏▏▏
Read: a 30-day chargeback rate is a down-payment on the truth, not the truth.
—
Ещё больше cpm benchmarks в @pixelprofit_fb
Median EPC sits at $0.41
EPC = earnings per click, gross revenue divided by raw clicks. Pulled across 40 networks, mixed verticals, 90-day window.
— P25: $0.18 ▁▁▁
— Median: $0.41 ▁▁▁▁▁▁
— P75: $0.94 ▁▁▁▁▁▁▁▁▁▁▁▁
— P90: $2.10 ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
The spread is the story: top-decile EPC runs ~5x the median (n=40). A single blended EPC number across offers is close to meaningless.
Read: judge EPC per-offer per-geo, never as a network average.
n=40 networks
EPC = earnings per click, gross revenue divided by raw clicks. Pulled across 40 networks, mixed verticals, 90-day window.
— P25: $0.18 ▁▁▁
— Median: $0.41 ▁▁▁▁▁▁
— P75: $0.94 ▁▁▁▁▁▁▁▁▁▁▁▁
— P90: $2.10 ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
The spread is the story: top-decile EPC runs ~5x the median (n=40). A single blended EPC number across offers is close to meaningless.
Read: judge EPC per-offer per-geo, never as a network average.
n=40 networks
Median hold time is 31 days
Hold = lag between conversion and the funds becoming withdrawable. Sampled from posted payment terms, 38 networks.
— ≤7 days: 11% of networks
— 8–30 days: 42%
— 31–60 days: 34%
— 60+ days: 13%
📊 ≤7d ▇▇
8-30 ▇▇▇▇▇▇▇▇
31-60 ▇▇▇▇▇▇
60+ ▇▇
Hold compounds with payment frequency: a 30-day hold plus net-15 terms can push first cash-out to ~45 days from the first conversion.
Read: model cash-flow on hold + frequency together, not the headline number.
n=38 networks
Hold = lag between conversion and the funds becoming withdrawable. Sampled from posted payment terms, 38 networks.
— ≤7 days: 11% of networks
— 8–30 days: 42%
— 31–60 days: 34%
— 60+ days: 13%
📊 ≤7d ▇▇
8-30 ▇▇▇▇▇▇▇▇
31-60 ▇▇▇▇▇▇
60+ ▇▇
Hold compounds with payment frequency: a 30-day hold plus net-15 terms can push first cash-out to ~45 days from the first conversion.
Read: model cash-flow on hold + frequency together, not the headline number.
n=38 networks
Chargeback rates span 0.4% to 6.1% by vertical
Chargeback rate = reversed transactions / approved transactions, trailing 6 months.
— Nutra (rebill): 6.1% ▇▇▇▇▇▇▇▇▇▇▇▇
— Sweepstakes: 3.2% ▇▇▇▇▇▇
— Dating: 1.9% ▇▇▇▇
— Finance lead: 0.9% ▇▇
— SaaS trial: 0.4% ▇
High-rebill verticals carry the reversal risk back to the affiliate via clawbacks, so a 6% chargeback line can erase a 15% margin.
Read: discount projected payout by the vertical's reversal rate before scaling spend.
n=22 advertisers
Chargeback rate = reversed transactions / approved transactions, trailing 6 months.
— Nutra (rebill): 6.1% ▇▇▇▇▇▇▇▇▇▇▇▇
— Sweepstakes: 3.2% ▇▇▇▇▇▇
— Dating: 1.9% ▇▇▇▇
— Finance lead: 0.9% ▇▇
— SaaS trial: 0.4% ▇
High-rebill verticals carry the reversal risk back to the affiliate via clawbacks, so a 6% chargeback line can erase a 15% margin.
Read: discount projected payout by the vertical's reversal rate before scaling spend.
n=22 advertisers