What Is Retention Rate? Cohorts, Curves, Benchmarks

Retention rate is the share of users still active N days after they started — the metric that separates products people use from products people tried. Acquisition fills the bucket; retention is whether the bucket holds water, and no growth spend survives a leaky one.

Cohort-based measurement (the only honest kind)

Take everyone who signed up in a given week (a cohort), then ask: what fraction was active 1, 7, 30 days later? Sitewide 'active users' blends new signups with veterans and hides everything; cohort curves show truth — each cohort's decay, comparable against its siblings. A product improving shows newer cohorts retaining better at the same age.

Reading the curve

• The early cliff (day 0–7): onboarding and activation territory. Most products lose most users here; improvements here compound through the whole curve.
• The flattening: every healthy curve plateaus — the users for whom the product became a habit. The plateau's height is your product-market fit, expressed as a percentage.
• No plateau: a curve sliding toward zero means a tool people finish with or a problem that expires. Fix the product story before the marketing.

Defining 'active' honestly

Logins flatter; value actions tell the truth. Define active as performing the core action (the same event your feature adoption setup tracks), or you will count zombie sessions — users checking whether they still need you — as healthy.

Day-1 / day-7 / day-30 framing

Consumer apps obsess over D1/D7/D30; B2B SaaS breathes weekly or monthly (nobody invoices on Saturday). Pick intervals matching your product's natural rhythm, then keep them fixed forever — retention numbers are only meaningful against their own history. Segmented by acquisition source, the same curves also grade your channels: the channel whose cohorts retain is the channel that deserves the budget, whatever its CPL says.