Apps lose about 39% of paying users every year, and the cancel flow is the one place most of them can still be argued with. Churnkey's team, working from data on 12 million people who tried to cancel, found that a well-built flow can get one in three to change their mind. The offer is the lever, but an offer only works when it matches the reason someone is leaving. Offers that win people back Discounts and pauses Discounts are the most commonly accepted offer. People who take one stay about five months longer, and one in ten are still paying a year later, even after the discount ends. The limit is price. Churnkey's guidance puts the sweet spot at 40% and under, because that range pairs the highest save rate with the highest customer lifetime value. The speaker is blunt about going deeper: 'And while it is high, the LTV is decimated by that.' Three-month discounts seem to perform best, and a discount stops working when it is offered every single time. Blocking coupon and pause stacking is the recommended guard against gaming, though that advice comes from the team rather than from measured results. Pauses look like the gentler offer. They keep customers about 5.5 months longer, yet only 7% of apps on Mobbin offer one. The risk is forgetting. A customer who forgets they paused meets a charge and files a dispute or cancels outright, which is why Churnkey recommends a maximum duration. Clout (up to 3 months, with the resume date shown), Hers (the next ship date), Disney Plus (a duration picker), and Juice Box (pause start and end stated) are the working examples offered. Plan switches and trial extensions Budget is the most common cancel reason, and matching it with steep discounts does not reliably produce the best results. The stronger move is a plan that fits. People who switch to a cheaper plan stayed about 7 months longer, and almost 30% were still around a year later. Spotify shows cheaper plans before the cancel button, Netflix offers an ad-supported plan, and Revolut offers one free month. Trial extensions answer a different fear. Memrise's founder said half of their cancellations happen within 3 minutes of starting a trial. People cancel right away because they do not trust themselves to remember before renewal. Duolingo's reminder two days before the trial ends helps, but a reminder only delays the decision. Opportune extends its trial by 3 months. Only 2% of Mobbin apps offer an extension at cancellation, so the tactic is rare, and the material does not establish how it performs across industries. When every offer is declined When the offers fail, the exit survey and the confirmation screen do the remaining work. Remind customers what they are giving up, but personalize the number. Foodpanda shows how much each customer has saved, and Wolt does the same, which works unless a customer has saved nothing. Canva shows older, cheaper pricing at the moment of cancelling. The speaker is unsure whether that reads as accountability or guilt-tripping, and the evidence does not settle it. Survey wording is where many flows lose customers. Only 18% of apps on Mobbin frame money as budget rather than 'too expensive', and 'taking a break' keeps a pause in play where 'I don't use it anymore' closes it. Churnkey's team recommends allowing one reason per respondent and removing the 'Other' option. The transcript reports that 20% of customers have accepted the other option, but the material does not say what that figure measures or how it compares with a flow without the option. Treat the removal as Churnkey's position rather than a settled result. Easy cancellation is itself a retention tactic. Save rates rise when the cancel button is easy to find and a manage-subscription option sits where customers expect it. A cancel button on every screen does not appear to lower saves, and the speaker argues it builds trust: 'You build a lot more trust with the customer if they have the ability to cancel on every screen.' Confirmation screens that name the customer, give the exact cancellation date, and promise no further charges appear in about 60% of Mobbin apps. Metrics that show the flow is working Churn is the share of subscribers who leave. 'If you lose 5% of people each month, almost half are gone by the end of the year.' Most companies Churnkey works with see mostly voluntary churn, meaning chosen cancellations rather than failed payments. Save rates of 20 to 36% are typical by industry, but a 50% save rate signals the offer is too high: 'We don't even really want to be saving 50% of customers.' Reactivation rates of 40 to 70% are considered healthy, and at 75% the flow is probably giving too much away. LTV decides whether saved customers are worth keeping. These benchmarks come from one practitioner's dataset, so measure your own before adopting them. The chapter's promise is a cancel flow that saves a real share of churners without making them feel trapped. Start by auditing each offer against the cancel reason it answers, cap discounts near 40% and set a pause limit, then track save rate and LTV together so the saves are worth keeping. Match each offer to its cancel reason: discounts for budget, pauses for non-usage, plan switches for price, and feedback paths for missing features. Keep discounts at 40% and under, avoid repeating them, and block coupon and pause stacking. Set a maximum pause duration and show the resume date so customers do not forget they paused. Frame money as budget and leaving as a break, allow one reason per respondent, and drop the 'Other' option. Track save rate and LTV together; a save rate near 50% suggests the offer is too generous.