Benchmark Watch: July 2026

Three new SaaS benchmark reports landed this quarter. KeyBanc published its 16th Annual Private SaaS Company Survey. High Alpha published its 2025 SaaS Benchmarks. Benchmarkit × Aleph published the 2026 SaaS and AI Performance Benchmarks.

CAC paybacks keep compressing

Benchmarkit × Aleph puts the 2026 median CAC payback at 16 months. That figure rests on 198 of the 342 companies surveyed, on 2025 actuals. The top quartile pays back in 6 months or faster, the bottom quartile past 24.

KeyBanc surveyed over 400 private SaaS companies with a median ARR around $45M. Its median is roughly 18 months on 2024 actuals, down from about 24 months in its 2023 edition. Its top quartile sits at 12 months. High Alpha’s cut for the $1M to $5M ARR bracket comes in lower, at an 8 month median.

Our baseline is a rolling meta-analysis of published benchmark studies: KeyBanc, High Alpha, Benchmarkit, AppsFlyer, and roughly fifteen others. The published ranges hold: 6 to 12 months for SMB and self-serve, 12 to 24 months for enterprise.

What moved is where you sit inside them. An 18 month payback read as healthy in 2022. Now it is average. Twelve months was once merely strong. Now it is the top quartile line for enterprise deals.

Benchmarkit × Pavilion’s 2025 dataset (1,600+ companies) puts median LTV:CAC at 3.6:1, mid range on our 3:1 to 5:1 range.

AppsFlyer’s State of Subscriptions 2026 report covers 1.7 billion paid installs and 2,900 apps. It puts cross category retention at 26% on Day-1, 11% on Day-7, and 5.4% on Day-30. That matches our consumer ranges.

A retention number that needs its definition checked first

Mixpanel’s State of Digital Analytics 2026 report is built from 577 billion B2B events. It publishes a North America “one-week retention” figure of 5.0% (APAC 8.3%).

Read on its own, that 5.0% sits inside our B2B Day-7 cohort range of 4% to 20% (an estimate). It sits far below the 40% to 60% returning-user rate from Pendo. Read the definition before you compare.

The same report publishes a separate “weekly retention” metric at 44.6% to 77.9%. Two metrics with near identical names measuring different things is a known trap in analytics reporting.

The consensus baseline uses a day-7 cohort measure. We have not yet confirmed that “one-week retention” means the same thing. Until we have, we hold our number.

The consensus baseline holds

Nothing in this scan changes our published ranges.

See where your own numbers land against the consensus baseline. Check your numbers against the baseline → benchmark.scilla.studio. Six minutes, no signup.