Founder retention

Purpose

Make repeated user value visible through correctly defined cohorts and connect every retention movement to a segment, lifecycle failure, and product action.

Give the founder a decision, the evidence behind it, the strongest contrary case, and the smallest executable next step. Do not produce generic encouragement or a long menu of tactics.

When to Trigger

Use this skill according to the frontmatter description. If the stated issue is a symptom of an earlier broken link, say so and route to founder-debugger or the owning retained skill.

Inputs Needed

• Target segment and acquisition source
• Core value event and its raw instrumentation
• Natural repeat cadence
• Signup, activation, core action, payment, expansion, contraction, and churn events
• User, account, logo, revenue, usage, or GMV unit
• Cohort start rule and observation window
• Customer contract length and cancellation freedom
• Qualitative evidence from retained, churned, and failed-activation users

Ask only for missing inputs capable of changing the decision. For reversible actions, make assumptions explicit and propose a bounded test instead of blocking on perfect data.

Questions to Ask

• What action proves the user received value?
• How often should a successful user naturally repeat it?
• Who belongs in the denominator?
• Does the curve flatten for any segment?
• Are newer cohorts improving?
• Is revenue retention hiding logo churn or vice versa?
• Are annual contracts masking weak usage?
• Did acquisition mix change?
• Where does the lifecycle first diverge between retained and churned users?
• What do the best retained users do before everyone else?
• Is involuntary churn separated from lost value?

Mental Models

• Retention is repeated value, not login frequency.
• Natural cadence before benchmark.
• Cohort before aggregate.
• Core-action retention before generic activity.
• Activation is the first retention intervention.
• Flattening reveals a retained segment, not universal PMF.
• Logo, revenue, usage, and GMV retention answer different questions.
• Annual contracts require engagement evidence.
• Cost per retained customer connects retention to distribution.

Use these models as competing lenses. Select the one that best explains the observed behavior, state what evidence would falsify it, and convert it into a decision rather than repeating it as a slogan.

YC Principles

• Repeat usage is stronger evidence than what users say.
• Choose the cadence from the product: daily, weekly, monthly, annual, or event-driven.
• A flattening cohort curve indicates PMF only for that product and retained segment.
• Do not scale growth channels while users flow out of the product.
• Track the actual core action, not vanity activity.
• Talk to retained and churned users to understand the mechanism behind the curve.

Paul Graham Principles

• Make a small number of users intensely happy before broad acquisition.
• Contact with users should change the product toward what they repeatedly need.
• Growth without strong appeal can conceal a moderately wanted product and create the fatal pinch.

Garry Tan Principles

• Treat retention as a product-craft outcome, not a lifecycle-messaging project alone.
• Use AI to analyze cohorts and qualitative evidence while preserving stable event definitions.
• In AI products, separate tourists from durable workload users and monitor dependency on underlying model changes.

Decision Frameworks

Metric selector

1. Consumer habit: user core-action retention at daily or weekly cadence.
2. Infrequent consumer job: event or annual return at the natural opportunity window.
3. B2B SaaS: account/logo retention plus active workflow and gross/net revenue retention.
4. Usage product or API: retained active accounts and usage/revenue curves.
5. Marketplace: buyer, seller, transaction, and GMV retention by market.
6. Annual contracts: usage, breadth, depth, and outcome health before renewal.
7. AI: task or workflow retention after tourist cohorts wash out, plus cost and correction rate.

Churn decomposition

1. Bad-fit acquisition
2. Failure to activate
3. Core value not delivered
4. Value delivered too rarely
5. Reliability or trust failure
6. Price-value mismatch
7. Customer business failure or champion loss
8. Competitive switch
9. Involuntary payment failure
10. Seasonal or natural dormancy

Intervention order

1. Verify instrumentation.
2. Segment the curve.
3. Fix acquisition quality if bad-fit users dominate.
4. Fix activation if retained users share an early behavior others miss.
5. Fix core product if activated users still do not repeat.
6. Fix reliability and trust before reminders.
7. Use lifecycle communication only when value exists but triggers are missed.
8. Scale acquisition only after retained cohorts justify it.

Step-by-Step Process

1. Define the user, core action, natural cadence, cohort start, and denominator.
2. Audit raw events and identity/account stitching.
3. Plot cohorts by segment and acquisition source.
4. Compare user, account, revenue, usage, or GMV retention as appropriate.
5. Choose the business object that can actually churn: use account retention when an account buys and renews, user retention when individuals independently receive value, and report both when seat depth predicts renewal.
6. Treat small cohorts as directional evidence: show counts beside percentages, compare multiple cohorts, and avoid scaling from one apparent plateau until the mechanism repeats.
7. Find the first lifecycle divergence between retained and churned users.
8. Interview and observe both groups.
9. Classify churn mechanisms.
10. Choose the earliest controllable mechanism.
11. Ship one intervention and annotate the cohort.
12. When the natural retention cadence exceeds the decision window, use an explicitly labeled leading indicator tied to the known mechanism, but keep the later cohort readout scheduled; never relabel activation or stated intent as retention.
13. Wait the minimum honest observation window.
14. Compare newer cohorts and guardrail metrics.
15. Decide whether to iterate product, narrow segment, change acquisition, adjust price, or scale.

Checklists

Definition

• Core action represents value.
• Cadence matches the job.
• Cohort and denominator are explicit.
• Identity and event instrumentation reconcile.
• Segments and channels are separable.

Interpretation

• Curve shape and newer-cohort trend are shown.
• Contracts are not mistaken for engagement.
• Logo and revenue views are compared.
• Qualitative mechanisms support the numbers.
• Acquisition scaling is gated on retained quality.

Red Flags

• Retention means app opens or logins with no value action.
• DAU/MAU is used for an infrequent product.
• Only aggregate active users are shown.
• Annual contracts produce false confidence.
• A curve flattening near zero is celebrated without segment size or economics.
• Expansion from a few accounts hides widespread logo churn.
• The team sends reminders before fixing missing value.
• Acquisition mix changes but cohorts are not segmented.
• AI model changes alter quality without cohort annotation.

Common Mistakes

• Using a universal retention benchmark.
• Changing cohort definitions between reports.
• Ignoring failed activation.
• Blending free, paid, consumer, SMB, and enterprise users.
• Treating cancellation reasons as causal truth without behavior.
• Optimizing resurrection while new cohorts remain broken.
• Calculating LTV from an unflattened curve.
• Scaling GTM based on bookings rather than retained customers.

Metrics

• Core-action retention curve
• Activation into first value
• Time to first value
• Logo/account retention
• Gross revenue retention
• Net revenue retention
• Usage or GMV retention
• Expansion, contraction, and resurrection
• Involuntary versus voluntary churn
• Cost per retained customer
• Retention by segment and channel
• Newer-cohort improvement
• AI task success, correction, and cost among retained users

For every metric, define the unit, numerator, denominator, cohort, segment, cadence, source event, and owner. Prefer decision thresholds and cohort movement over universal benchmarks.

Example Scenarios

Scenario 1

Travel users book twice a year: use opportunity or annual booking retention, not DAU/MAU; measure whether users return when the job recurs.

Scenario 2

Annual enterprise contracts renew but usage collapses: treat workflow activity and realized outcome as leading retention; contract lock-in is not PMF.

Scenario 3

An AI app has large month-one churn: rebase analysis after tourist cohorts, isolate retained workloads, and compare task success, frequency, willingness to pay, and model dependency.

AI Prompt Templates

Template 1

Define the correct retention metric for this product, including unit, core action, cohort start, cadence, denominator, segments, and observation window.
End with a decision, owner, deadline, metric, threshold, stop rule, and strongest contrary case.

Template 2

Diagnose these cohort curves. Separate instrumentation, acquisition quality, activation, core value, reliability, pricing, and customer-context causes.
End with a decision, owner, deadline, metric, threshold, stop rule, and strongest contrary case.

Template 3

Design one retention experiment from this lifecycle divergence with owner, event changes, target cohort, threshold, guardrails, and honest observation period.
End with a decision, owner, deadline, metric, threshold, stop rule, and strongest contrary case.

Related Skills

• founder-talking-to-users
• founder-mvp
• founder-distribution
• founder-pricing
• founder-default-alive
• founder-debugger

Further Reading

1. Startup School Week 4: Launching and Growth — Links manual launch, retention, funnels, product growth, and external channels.
2. Growth Office Hours — How to derive the core action and natural retention interval from retained-user behavior.
3. David Lieb on Google Photos and Bump — A concrete postmortem on why downloads and actives can conceal misunderstood retention.
4. Analytics for Startups — Instrumentation and product analytics for the PMF journey.
5. The Real Product-Market Fit — Distinguishes a decent business or product from unmistakable market pull.
6. Growth AMA with Gustaf Alströmer — Stage-specific channel selection, founder-led growth, retention prerequisites, and marketplace sequencing.

Source Links

• Startup School Week 4: Launching and Growth — Kat Mañalac and Gustaf Alströmer. Links manual launch, retention, funnels, product growth, and external channels.
• Growth Office Hours — Anu Hariharan and Gustaf Alströmer. How to derive the core action and natural retention interval from retained-user behavior.
• David Lieb on Google Photos and Bump — David Lieb and Gustaf Alströmer. A concrete postmortem on why downloads and actives can conceal misunderstood retention.
• Analytics for Startups — Y Combinator. Instrumentation and product analytics for the PMF journey.
• The Real Product-Market Fit — Michael Seibel. Distinguishes a decent business or product from unmistakable market pull.
• Growth AMA with Gustaf Alströmer — Gustaf Alströmer. Stage-specific channel selection, founder-led growth, retention prerequisites, and marketplace sequencing.
• Retention Is All You Need — Andreessen Horowitz. Secondary analysis of AI-tourist churn, curve flattening, engagement, and cost per retained customer.
• 16 More Startup Metrics — Andreessen Horowitz. A specific cohort-analysis procedure and warnings about vanity metrics.
• How to Know If You've Got Product-Market Fit — Lenny Rachitsky. Secondary synthesis emphasizing long-term cohort retention and market-specific flattening.
• What I've Learned from Users — Paul Graham. Tests whether founders are paying attention by asking what, specifically, they have learned from users.