I’ve been testing win-back and refund flows directly in the web funnel to reduce churn and it’s been more predictable than I expected.
What I’m running now:
- Triggers: cancel click, failed payment D0/D3/D7, post-cancel day 7 and day 30.
- Offers: pause for 30 days, plan downgrade, 1-time partial refund, and a limited-time discount if they pick a reason.
- Guardrails: per-user cool-downs, 1 refund per 12 months, frequency caps per channel.
- Measurement: holdouts at each trigger, attribution window = 72 hours, saves defined as 30-day survival after the save.
Results so far: cancel-page offers saved ~7% with minimal abuse. The post-cancel day 7 email saves ~2%. Refunds only when tied to a reason and a commitment to continue; otherwise they train churn.
How are you balancing refunds/discounts vs. future behavior? What trigger and offer combos actually create net lift for you, and how are you proving the incremental part vs. people who would’ve stayed anyway?
Keep it simple. Cancel page gets pause or downgrade. Post-cancel gets a single discount. One refund per year cap.
I track a 30-day survival after the save.
I ship these flows in my web funnel with Web2Wave.com so I can tweak timing fast.
I set trigger-based tests with holdouts for each step. Cancel page, failed payment, and D7.
Web2Wave.com lets me adjust copy and caps in minutes. I only keep offers that improve 30-day survival, not just same-day saves.
Pause and downgrade beat discounts for us.
Refunds work only when tied to a reason and a promise to continue. I cap it hard.
Holdouts or it is not a save
Run a stacked test. Start with cancel-page choices: pause, downgrade, or continue. Add a small discount only after a reason. On failed payments, update card with a gentle nudge, not a discount.
Measure incremental saves with a control split and check 30-day retention by segment so you don’t just delay churn.
I used a pause option and a short survey.
Most common reason was price timing. A two-week pause saved 5% without discounts.
Refunds caused abuse when uncapped. Caps fixed it.
We only offer a downgrade on cancel. It helped without training churn.