A telecom ran the holdout test almost nobody runs. Their retention campaign was causing 2% more churn — not less. Here's how that happens. Every save flow tracks one number: save rate — the share of people who start cancelling and don't finish. A 30% save rate sounds like a win. But it hides four completely different users: → Persuadables — will leave unless you intervene. The only ones a save flow actually helps. → Sure Things — were never leaving; that discount was wasted margin. → Lost Causes — already decided; a 4-page flow just makes them angry, and looks a lot like the cancel maze that cost Amazon $2.5B. → Sleeping Dogs — quietly renewing, forgotten they're paying. Your "We'd hate to see you go" email just reminded them. Sleeping Dogs are the trap. When your save flow contacts everyone, it nudges people who were going to keep paying toward the exit. For a 200K-subscriber business, that's roughly $360K–$720K a year in churn your own retention system manufactured. Most companies never catch it, because they never run the holdout test that telecom did. The fix isn't a better offer. It's knowing who's a Persuadable and who's a Sleeping Dog before you touch them — inside a cancel flow that survives the FTC's bright line: if cancelling takes more steps than signing up, you have exposure, no specific dark pattern required. So what is your save rate really measuring — saves, or Sleeping Dogs you scared into leaving? #SubscriptionRetention #ChurnReduction