Churn basics

Churn Basics

At small scale, churn is not a dashboard problem — it's a "talk to the five people who left" problem. Acquisition fills the bucket; this skill patches the holes. It's for products with real users (even 10) where signups aren't turning into a growing base.

Context

Read product-context.md if present. Otherwise ask for:

1. Product, pricing model, and what the core value action is
2. Current numbers: signups/mo, active users, paying users, cancellations
3. What analytics exist (events table, PostHog, Stripe) — even rough
4. Any known complaints or cancellation reasons so far

Process

1. Measure it honestly (30 minutes)

Two numbers, monthly:

• Customer churn = customers lost in month ÷ customers at start of month
• Activation-adjusted retention — of users who signed up 30 days ago, how many did the core action in the last 7 days?

Early-stage reality check: 5–10% monthly customer churn is normal for young SaaS; >15% means the product isn't delivering its promise yet. With < 50 customers, ignore percentage noise — count people and talk to each one.

2. Find the drop point

Users churn at predictable cliffs — locate which one is yours:

Cliff	Symptom	Likely cause
Never activated	Signed up, never did the core action	Onboarding / unclear first step
Used once, gone	Did it once, never returned	Value was a novelty, not a workflow
Gone at first invoice	Cancelled when billing hit	Price ≠ perceived ongoing value
Slow fade	Usage decayed over weeks	Product lost to habit competition

Pull the last 10 churned users and tag each with a cliff. The most common tag is the work queue.

3. Exit interviews that get answers

Email every cancellation within 24 hours. Three rules: one question, reply-to goes to the founder, no survey link.

Subject: quick question (no win-back attempt)

Saw you cancelled [Product] — totally fine, and I won't try to talk you out of it.
One question so I can fix it for the next person:
what was the moment you decided it wasn't worth keeping?

— [Name], founder

30–50% reply to this when it's genuinely one question. "What made you decide" gets the trigger moment; "why did you cancel" gets a polite lie.

4. Ship the three standard fixes

Match the fix to the cliff — in leverage order:

1. Never activated → cut onboarding to one step that reaches the core action in < 3 minutes (see onboarding-activation); send a founder email on day 1 offering to set them up personally
2. Used once → build the return trigger: a weekly email/notification tied to their data ("your launch video got 400 views"), not a generic digest
3. Invoice churn → move the plan boundary so the invoice lands after a value moment, add pause-instead-of-cancel, and show usage on the billing page ("you made 12 videos this month")

Re-measure 30 days after each fix. One fix at a time — three at once tells you nothing about what worked.

Output

1. Churn snapshot — the two metrics computed from this product's actual data (or the SQL/queries to get them)
2. Cliff diagnosis — last 10 churned users tagged, with the dominant cliff named
3. Exit-interview email — personalized, ready to send on the next cancellation (+ automation note for Stripe webhook → email)
4. Fix plan — the one fix matched to the dominant cliff, specced concretely for this product, with the 30-day re-measure date
5. Retention dashboard note — where these two numbers live from now on (KPI script, PostHog insight, or a pinned SQL query)

---

Retention starts before signup — the right users churn less. SoloMax launch videos show your product doing the real thing, so the users who arrive are the ones the product is actually for.