Retention

You are growing, but users are leaving as fast as they arrive. Retention is the metric that determines whether your growth is real or an illusion — whether you are filling a bucket or pouring water into a sieve. Every growth channel, every marketing campaign, every product improvement is wasted if users do not stay. Retention is not one of many metrics. It is the metric that validates all the others.

The Core Idea

Retention measures the percentage of users who return to your product over time. A retention rate of 40% at day 30 means that 40 out of every 100 new users are still active one month after signing up. The shape of the retention curve — not just the absolute number — tells you whether your product has found product-market fit.

Naomi Gleit, head of product at Meta and Facebook’s 29th employee, describes the foundational insight that created the legendary Facebook growth team: “What we found was the churn and resurrection lines were actually much larger than the new user line, which implied to us that retention and driving those two lines was actually our biggest lever to drive net growth.” The growth accounting equation — new users minus stale users plus resurrected users — revealed that focusing on acquisition was solving the wrong problem. The bigger lever was keeping people who had already signed up. This insight redirected Facebook’s entire growth strategy from acquisition to retention.

Reading Retention Curves

The retention curve is a plot of active users (y-axis) over time since sign-up (x-axis). Every product’s retention curve eventually flattens or hits zero. The question is where it flattens.

The Three Retention Curve Shapes

Shape	What It Means	Implication
Flattening curve (levels off above zero)	A segment of users finds ongoing value	You have PMF for that segment. Focus on expanding it.
Declining curve (always trending down)	No segment retains durably	You do not have PMF yet. Fix the product before scaling acquisition.
Smiling curve (dips then recovers)	Users who leave come back	Seasonal products, reactivation working, or habit formation delayed

A flattening retention curve is the most reliable signal of product-market fit. Sarah Tavel, partner at Benchmark and former first PM at Pinterest, frames retention through her Hierarchy of Engagement: “The test for me, of whether you’re building a product that has the ingredients to create a retentive product, is that the product should get better the more you use it, and you’ll have more to lose by leaving it.” At Pinterest, this meant that the more you pinned, the more personalized your feed became and the more you had invested in your boards — creating compounding value that made leaving increasingly costly.

Retention Windows

Different products measure retention on different time scales:

Product Type	Primary Retention Window	Why
Social / messaging	D1, D7, D30	Daily use expected
B2B SaaS	Weekly, Monthly	Work-driven cadence
E-commerce / marketplace	Monthly, Quarterly	Purchase frequency
Consumer subscription	Monthly, Annual renewal	Billing cycle
Seasonal (tax, travel)	Annual	Usage is seasonal by nature

Measuring retention on the wrong window produces misleading data. A B2B tool measured on D1 retention will look terrible — users do not use work tools on Saturday.

Diagnosing Retention Problems

New User Retention (Day 0-30)

If users leave in the first week, the problem is almost always onboarding and activation:

Symptom	Likely Cause	Fix
Massive drop-off Day 0-1	Product does not deliver value quickly enough	Reduce time to value, improve first-run experience
Drop-off between D1 and D7	Users activated but did not form a habit	Introduce habit-forming loops, trigger-action patterns
Drop-off between D7 and D30	Initial value wore off, no deeper engagement	Reveal additional value layers, drive feature discovery

Naomi Gleit describes how Facebook tackled early retention by identifying the variable most correlated with it: “How do we drive engagement and retention? We look at the variables that correlate most with that outcome. What we found was friending.” The famous “seven friends in 10 days” activation metric was not about a magic number — it was about reaching a threshold where the product became compelling. As Gleit notes: “Facebook is much more compelling if you have 14 friends. And the other thing around 10 or 14 days is we wanted it to happen quickly, we wanted to have you experience the magic moment soon after you had registered on the site to prevent you from churning.”

Medium-Term Retention (Month 1-6)

If users leave after the first month, the problem is typically value exhaustion or competitive alternatives:

Symptom	Likely Cause	Fix
Steady monthly decline	Users extracted the initial value but found no ongoing need	Build ongoing use cases, content refresh, social features
Sudden drop at a specific point	Feature limitation, pricing wall, or alternative discovered	Investigate what happens at that time threshold
Segment-specific decline	Certain user types do not retain	Focus on best-retaining segments, deprioritize poor-fit segments

Long-Term Retention (6+ months)

If established users leave, the causes are usually:

• Job change (B2B — the champion leaves the company)
• Competitive substitution (a better alternative emerges)
• Value saturation (they have gotten what they needed)
• Quality degradation (the product got worse — bugs, bloat, price increases)

Retention Benchmarks

Approximate benchmarks drawn from patterns across multiple episodes:

Product Type	D1	D7	D30	Good?
Top consumer social	60%+	40%+	25%+	Top quartile
Average consumer app	35%	15%	8%	Median
B2B SaaS (good)	N/A	80%+ weekly	70%+ monthly	Strong
B2B SaaS (average)	N/A	60% weekly	50% monthly	Adequate
Marketplace (good)	N/A	N/A	30%+ monthly	Transaction-based

The most important comparison is not to benchmarks but to your own cohort trends. Is each new cohort retaining better than the previous one? If yes, the product is improving. If not, something is degrading.

Cohort Analysis

Raw retention numbers can be misleading because they blend early and late users. Cohort analysis separates users by sign-up date and tracks each cohort independently.

Why cohorts matter:

A product with 50% overall monthly retention might consist of:

• Old cohorts retaining at 70% (loyal users)
• Recent cohorts retaining at 20% (acquisition channel brought low-intent users)

The aggregate number hides that recent growth quality is terrible. Cohort analysis reveals this.

What to Look For

Pattern	What It Means
Each newer cohort retains better	Product is improving; changes are working
Each newer cohort retains worse	Acquisition quality declining, or product changes hurting new users
All cohorts flatten at the same level	Stable product with consistent value proposition
Old cohorts declining after being flat	Something changed — investigate feature changes, competitor entry, or degradation

Improving Retention

Bangaly Kaba, who was an early growth PM at Facebook, head of growth at Instagram, and VP of product at Instacart, warns against the most common retention improvement anti-pattern: “Someone says, ‘Hey, you know what? This would be great to build.’ And you go pull data to go justify why that would be great to build. Call that identify, justify, execute.” His alternative: “First you have to really understand from first principles what is actually going on. So understand, identify, execute.” The understand step — instrumenting data, watching user behavior, mapping drop-off points — is where most retention insights come from. Skipping it leads to building solutions for problems that do not exist.

The Retention Equation

Retention = Value Received > Effort Required + Alternatives Available

To improve retention, you can:

1. Increase value delivered — More features, better experience, more content
2. Reduce effort required — Better UX, automation, fewer friction points
3. Increase switching costs — Data lock-in, integrations, social graph, workflow investment (but do not manufacture artificial lock-in — it breeds resentment)

Tactical Retention Levers

Lever	How It Works	Example
Habit loops	Product use becomes automatic through trigger-action-reward patterns	Duolingo streak counter
Content freshness	New content gives users a reason to return	Netflix new releases
Social obligation	Users return because others depend on them	Slack (team expects response)
Investment effects	The more the user puts in, the more they have to lose by leaving	Notion (accumulated knowledge), Figma (team files)
Lifecycle emails	Re-engage users who are drifting away	”You have 3 unread messages”
Feature discovery	Surface features users have not tried	”Did you know you can…?”

Sarah Tavel extends retention thinking to its ultimate form — self-perpetuating loops: “Every time a user uses your product, I love to think of it as this kinetic energy that they’re putting into your product. Your job with a great product is to take that energy and, as much as possible, convert it back to the experience that they’re having.” Products that fail to create these loops, like Evernote (no collaborative features, no network effects), eventually tap out on growth because they must pay to acquire every new user.

Retention and Growth: The Compounding Effect

Here is why retention is the most important growth metric:

A product with 10,000 monthly sign-ups and 50% monthly retention:

• Month 1: 10,000 active users
• Month 6: ~31,000 active users
• Month 12: ~39,000 active users

The same product with 60% monthly retention:

• Month 1: 10,000 active users
• Month 6: ~49,000 active users
• Month 12: ~88,000 active users

A 10 percentage point improvement in retention more than doubled the active user base at month 12 — with zero change in acquisition. This is the arithmetic behind Naomi Gleit’s insight at Facebook: the churn and resurrection lines were larger than the new user line, making retention the highest-leverage growth investment.

Key Takeaway

• Retention is the metric that validates all others. Growth without retention is waste.
• Read retention curves, not just numbers. A flattening curve is the strongest signal of product-market fit.
• Analyze retention by cohort, not in aggregate. Aggregate numbers hide whether product quality is improving or degrading.
• Fix new-user retention (onboarding and activation) first — it is where the largest volume of users is lost.
• A small improvement in retention compounds dramatically over time, often exceeding the impact of doubling acquisition spend.

Related

• Activation Rate — Activation is the leading indicator of early retention
• Onboarding — The experience that determines first-week retention
• North Star Metric — Retention is typically a key input in the NSM tree
• Product-Market Fit — Flattening retention curves are the strongest PMF signal
• Growth Loops — Retention determines whether loops compound or decay

Sources

• Naomi Gleit on Meta growth — Growth accounting framework, 7 friends in 10 days, shift from acquisition to retention, understand-identify-execute
• Sarah Tavel on the hierarchy of engagement — Core action framework, accruing benefits and mounting losses, self-perpetuating loops, Evernote vs Pinterest
• Bangaly Kaba on growth frameworks — Understand work, identify-justify-execute anti-pattern, onboarding-to-habit gap
• Sean Ellis on growth hacking — PMF test as retention predictor, onboarding to the right user experience
• Archie Abrams on Shopify growth — Optimizing for churn, long-term experiment holdouts on retention