The most common retention mistake I see: teams trying to fix everything at once. They build a save flow, a health score, an email sequence, and a referral program in parallel. Six months later they have four half-finished projects and the same churn rate.
The teams that actually move churn down do one thing at a time, in the right order. Here's that order.
(If you want a personalized version of this sequence based on your current setup, take the Churn Health Check. It asks 8 questions and tells you exactly which step to start with.)
Step 1 (always): Fix involuntary churn
Time to impact: 1-4 weeks
Difficulty: Easy
ROI: Highest of any retention investment
This is non-negotiable. 20-40% of your total churn is failed payments, and most of it is recoverable with proper dunning. The math is brutal: if you're not doing AI dunning, you're losing 5-15% of MRR per year to expired credit cards.
The fix takes a day to implement: turn on Stripe Smart Retries (free) or sign up for Churnkey ($100-300/month). Both pay for themselves within the first week.
This step is so much higher ROI than anything else that it should always be first. Read the AI dunning guide for implementation details.
Step 2: Build a cancellation save flow
Time to impact: 2-6 weeks
Difficulty: Easy-Medium
ROI: Very high
Your cancellation page is doing nothing right now. You can change that with 1-2 weeks of work and recover 15-25% of cancellation attempts.
The basic version: when a customer clicks cancel, ask why, then offer a save based on their reason. Price-sensitive? Discount or downgrade. Low usage? Pause option. Missing feature? Roadmap preview. The AI save flow guide has the exact framework.
Why this is step 2: it directly captures users about to leave, before they even leave. It compounds with step 1 because saved customers stay in your dunning recovery pool too.
Step 3: Build a health score (even a simple one)
Time to impact: 4-8 weeks
Difficulty: Medium
ROI: High
Now you can recover failed payments and rescue cancellations. The next move is to identify at-risk customers before either of those happens. That's a health score.
Don't overthink this. Start with rules: usage dropped 40%+ over 2 weeks + last login over 7 days ago + support ticket in the last 30 days = red. That simple rule will catch 60-70% of churning customers 30+ days early. The fancy ML version comes later.
Why this is step 3: it gives your CSM team something to act on. Without a health score, they're firefighting. With one, they're working a queue prioritized by risk.
Step 4: Fix activation/onboarding
Time to impact: 2-3 months
Difficulty: Medium-Hard
ROI: Compounding
Onboarding is the highest long-term lever, but it's also the hardest to change quickly because it requires product work. Steps 1-3 don't need product changes, just billing and CS workflows. Step 4 needs engineering.
Define your aha moment using the activation audit, then build personalized onboarding paths for different user segments. AI-personalized onboarding can cut first-month churn in half.
Why this is step 4: it touches 100% of new signups, so even small improvements compound massively. But it's slower to implement than 1-3, so it goes here.
Step 5: Build expansion revenue
Time to impact: 3-6 months
Difficulty: Hard
ROI: Strategic
By now you've reduced churn meaningfully. Time to flip the equation. Net Revenue Retention above 110% is what separates good SaaS companies from great ones, and the way you get there is by growing existing accounts.
This means usage-based pricing components, feature tiers that customers grow into, multi-seat expansion paths, and proactive upgrade prompts based on usage signals. The product-led expansion experiment has the playbook.
Step 6: Predict instead of react
Time to impact: 6+ months
Difficulty: Hard
ROI: Strategic
Now you have systems for activation, engagement, retention, and expansion. The final step is moving from reactive to predictive: AI churn prediction, automated interventions, dynamic pricing, all working together.
This is where most SaaS companies stop, but the top 10% keep going. Retention becomes a moat at this stage, not just a metric.
Why this order works
Three principles drive this sequence:
- Speed matters. Fast wins build momentum and prove retention is worth investing in. Steps 1-3 show measurable results within 2 months.
- Compounding matters. Each step makes the next one work better. A save flow without dunning still loses customers to failed payments. A health score without a save flow has nothing to escalate to.
- Engineering complexity ramps up. Steps 1-3 are mostly billing/CS workflows, no product changes needed. Steps 4-6 require engineering investment. Front-loading the easy stuff lets you ship while engineering plans the bigger work.
Personalize the sequence to your situation
The 6 steps above are the default. Your specific situation might call for a different order. If your involuntary churn is already low (you're on Stripe with Smart Retries), you can skip step 1. If you have no expansion motion at all, you might need to do step 5 earlier.
The fastest way to get a personalized sequence: take the Churn Health Check. It identifies your weakest dimension and gives you the right starting step for your specific gaps. 60 seconds, no signup, instant result.