Revenue Cohort Analysis

Net revenue retention by acquisition cohort · 1,448 customers · 54 months · synthetic data

Cohort heatmap — % of initial ARR retained, by months since acquisition

CohortARR@0 $MM0M3M6M9M12M15M18M21M24M27M30M33M36
2022Q11.4100101100100969898989494999797
2022Q21.3100979195949481777878777373
2022Q31.8100989791929292919182757575
2022Q42.4100959494815957576059594949
2023Q12.3100969595868784847878807676
2023Q22.3100999684828387838479788181
2023Q32.61009897818076757371707070·
2023Q42.7100100969393888785838182··
2024Q12.2100928989899088817976···
2024Q23.01009898948786848181····
2024Q3 ◂6.410099959386666663·····
2024Q4 ◂5.9100968983756969······
2025Q13.1100991021009998·······
2025Q24.610095878885········
2025Q34.0100989895·········
2025Q44.31009694··········
2026Q14.210097···········
2026Q23.1100············
◂ 2024Q3–Q4: the promo-acquisition cohorts — and the two largest cohorts on the page at $6.4M and $5.9M of starting ARR. They churn off a cliff at first renewal: ~11 points below neighboring cohorts at M12, ~15 at M18, and falling. Small cohorts (2022Q4) wobble by noise; these are dollar-weighted and systematic. Blended NRR barely registered any of it — which is the argument for cohort-level analysis.

Layer cake — ARR by cohort year

$0M$25M$49M20222023202420252026 ■ 2022 ■ 2023 ■ 2024 ■ 2025 ■ 2026

CAC payback curves by channel — cumulative gross profit / CAC, empirical

0.0x1.0x2.0x3.0xCAC recovered (1.0x)M0M6M12M18M24M30M36M42 ● inbound (13mo) ● outbound (>window) ● partner (29mo) ● self_serve (3mo)
Curves are cumulative observed gross profit over total CAC, per acquisition channel — no extrapolation beyond the observation window. The formulaic LTV (ARPA × margin ÷ churn) is deliberately absent.

LTV : CAC by segment — observed to date

SegmentCustomersCAC total PaybackGP/CAC to date
smb668$2.5M11mo1.7x
mid market576$17.3M16mo1.5x
enterprise204$33.4M24mo1.3x