Retention Curve

retention curve · retention chart · cohort retention curve · retention graph · survival curve · retention plateau

Definition

A retention curve is a time series of the share of a cohort that remains active; it shows the speed of churn and whether a loyal core of the audience exists.

How to read a retention curve

A retention curve is built for a cohort of users: the X axis holds the time since the first action
(a purchase, a registration), the Y axis the share of users who have repeated the action by that
point.

Every curve starts at 100% at point zero (everyone in the cohort is “alive”) and falls from there.
The shape of that fall says a great deal about the product:

Shape of the curve	What it means	What to do
Sharp fall → plateau	A loyal core; the product forms a habit	Raise the height of the plateau
Slow, steady decline	Gradual organic attrition	Find a trigger that brings people back
Fall to zero	No retention — the product is single-use	Rethink the value proposition
A bump at 30–60 days	Seasonality or a need cycle	Reflect it in the communication plan

The two key parameters of the curve

The speed of the initial fall — how sharply the curve drops in the first 7–14 days. This is the
moment of truth: the user has had the experience and decided whether to come back. In e-commerce the
first 30 days are critical, and that is exactly where personalization, relevant recommendations and
post-purchase triggers have the largest effect.

The height of the plateau — the percentage of users who stabilise after 90–180 days. This is the
durable loyal segment. Even a small upward shift of 2–3 percentage points changes the LTV of the
whole base substantially.

Cohort: 10,000 customers
Day 0:   10,000 (100%)
Day 30:   2,800 (28%)
Day 90:   1,400 (14%)
Day 180:  1,100 (11%)  <- plateau
Day 365:  1,050 (10.5%) <- stable

How to compare cohorts

The value of the curve lies not in the absolute number but in the comparison between cohorts. If a
cohort that arrived after personalization was deployed retains 3 percentage points better at 90 days,
that is a measurable effect of the programme.

Tip: when comparing cohorts, make sure they are comparable by acquisition channel. Organic
users retain better than performance traffic — mixing them distorts the result.

Retention curves in e-commerce vs SaaS

In SaaS the retention curve is built on weekly or monthly use of the application. In e-commerce the
definition of “active” needs a decision: does the return point mean a purchase, a visit or an add to
cart?

For retailers with infrequent purchases (electronics, furniture) it makes sense to count a site
visit rather than a transaction — otherwise the curve looks worse than reality. What matters is
fixing the definition and not changing it.

Frequently asked questions

How do you build a retention curve?

Take a cohort (for example, everyone who bought in January) and fix their first purchase as day 0. Then calculate what share repeated the target action after 7, 14, 30, 60, 90 and 180 days. The result is a series of points joined into a curve.

What does the plateau on a retention curve mean?

The plateau is the point where the curve stops falling and stabilises. It is the core of loyal users who stay with the product regardless of external factors. The height of the plateau is the key measure of product health. If no plateau forms and the curve runs to zero, the product is not creating a habit of returning.

Why is cohort analysis better than an average retention rate?

An average retention rate blends every cohort together. Cohort analysis lets you compare cohorts from different periods and judge whether retention improved after a specific change — new personalization, a redesign, a shift in assortment.