Churn and retention

About 51 episodes cover retention, though many only report a churn or retention figure. The strongest shared view is that retention is decided by the product and the problem, not by retention tricks: build something people need daily and that makes them more money than they pay, and output quality keeps them. Several founders say to secure retention before pushing acquisition or monetisation, since traffic sent to a leaky product mostly drains away. The most concrete levers are choosing the right customers (higher-paying, enterprise or high-intent customers churn less, and some reposition around the ‘magnet’ users who never leave), talking directly to churned or unhappy users, adding yearly plans, and building habit or activation features (a commitment screen, refining results, AI agents doing the work automatically). Reported churn ranges from roughly 7% to over 30% and depends heavily on category: job search, productised services and ghostwriting agencies churn fast by nature. Make cancelling easy rather than trapping users.

Key points

• Put retention before acquisition and monetisation: fix stickiness first, then push growth — 7 episodes. (Tibo · Tibo (Dec 2025) ▶ 10:41) (Ken · Ken (Sep 2026) ▶ 13:06) (Guillaume Moubeche · Guillaume Moubeche (Dec 2024) ▶ 17:16) +4 more
• Customer fit predicts churn: higher-paying, enterprise and high-intent customers stay; position around the users who never leave — 7 episodes. (unnamed guest · Josef Buettgen & Timo Nikolai (Sep 2025) ▶ 10:01) (Denis Yurchak · Denis Yurchak (Oct 2025) ▶ 11:17) (Gaurav · Gaurav (Aug 2026) ▶ 08:38) +4 more
• Retention comes from the product and the problem (sticky need, output quality, results for the client), not from retention tricks — 6 episodes. (Jack Friks · Jack Friks (Mar 2025) ▶ 15:02) (Jure Sotosek · Jure Sotosek (Apr 2025) ▶ 13:44) (Ramon Berrios · Ramon Berrios & Blaine (Jun 2024) ▶ 08:48) +3 more
• Build habit and activation into the product: commitment screens, features that refine results, collectables, agents that work automatically — 6 episodes. (Michael · Michael (Jul 2026) ▶ 12:15) (Mauro Botanes · Mauro Botanes (Sep 2026) ▶ 12:20) (Lotts Ezeike · Lotts Ezeike (Nov 2025) ▶ 09:22) +3 more
• Churn is structurally high in some categories (job search, productised services, agencies, some AI tools) — 6 episodes. (Roman · Roman (Sep 2026) ▶ 02:20) (Hunter Hammonds · Hunter Hammonds (Sep 2023) ▶ 07:49) (Alex Lieberman · Alex Lieberman (Nov 2023) ▶ 10:48) +2 more
• Talk to churned and unhappy customers directly, and use onboarding calls or win-back emails — 4 episodes. (Dmytro Krasun · Dmytro Krasun (Nov 2024) ▶ 07:21) (Andre Haykal Jr. · Andre Haykal Jr. (Jul 2024) ▶ 14:55) (Guillaume Moubeche · Guillaume Moubeche (Dec 2024) ▶ 13:13) +1 more
• Yearly (or lifetime) commitments reduce churn — 3 episodes. (Pauline · Pauline (Oct 2024) ▶ 08:38) (Connor Burd · Connor Burd (Oct 2025) ▶ 10:34) (Umberto Mezzadra · Umberto Mezzadra (Mar 2026) ▶ 16:20)

Where guests agree

• Top-of-funnel success can hide a retention problem; high trial numbers with nobody staying signals a product issue. (unnamed guest · Eyal & Yahli (Nov 2025) ▶ 03:33) (Tibo · Tibo (Dec 2025) ▶ 10:57) (Shervin Shares · Shervin Shares (Apr 2023) ▶ 12:32)

Frameworks

• Three churn fixes — What cut her churn: a yearly plan, a free trial, and better output quality. (Pauline · Pauline (Oct 2024) ▶ 08:40)
  • Add a yearly plan
  • Offer a free trial
  • Improve service quality
• Magnet persona — Identify the customers who never churn and whose identity attracts others (Apple’s designers; for lemlist, sales reps), then position the product around them so buyers want to be associated with that persona. (Guillaume Moubeche · Guillaume Moubeche (Dec 2024) ▶ 17:30)
  • Study which customers never churn once they join
  • Pick the persona that attracts more customers (a ‘cool’ or aspirational group)
  • Position the product as the tool for that persona
• Three pillars of growth — Growth measured through three pillars: distribution (getting people to know about the product), conversion (getting them to pay) and retention (how long they keep using it). (David Park · David Park (Mar 2025) ▶ 02:04)
  • Distribution
  • Conversion
  • Retention

Tactics

• Ask churned customers one yes/no question — Review each cancelling customer’s profile, company, use case and stated reason, guess why they churned and email one yes/no question (e.g. ‘did you churn because of pricing?’). (Dmytro Krasun · Dmytro Krasun (Nov 2024) ▶ 07:21)
• Fix churn by changing marketing copy — When wrong-fit customers (expecting no-code) churned, he emphasised that the product is for developers who write code to integrate it. (Dmytro Krasun · Dmytro Krasun (Nov 2024) ▶ 08:10)
• Offer every customer a one-on-one onboarding call — Ask why they signed up and what they hoped to achieve, check whether they’re getting it, and fix product or support gaps to keep churn low. (Andre Haykal Jr. · Andre Haykal Jr. (Jul 2024) ▶ 14:46)
• Call unhappy users directly after a risky change — After a redesign angered customers, he invited every unhappy user onto Zoom calls, working until 4am, found features they had wrongly removed and fixed them. (Guillaume Moubeche · Guillaume Moubeche (Dec 2024) ▶ 13:13)
• Find your non-churning customers and position around them — Look at who never churns after joining; for lemlist it was sales reps, which made it ‘the sales tool for sales teams’ in people’s minds. (Guillaume Moubeche · Guillaume Moubeche (Dec 2024) ▶ 17:27)
• Win back churned users by email — Use Loops to email churned users and convert them back. (Benji · Benji (May 2026) ▶ 05:13)
• Add a commitment screen at the end of onboarding — One screen where the user commits to working out raised retention by 3%. (Mauro Botanes · Mauro Botanes (Sep 2026) ▶ 12:20)
• Keep users subscribed with habit features that refine the result — Stretches, meal scanning and sleep logging update the prediction as the weeks go by. (Michael · Michael (Jul 2026) ▶ 12:15)
• Lead with one core message, then retain with a bundle — Market as the best AI carousel generator; offer extra tools inside so users stay. (Fernando Pessagno · Fernando Pessagno (Dec 2024) ▶ 09:28)
• Add a yearly plan to reduce churn — Yearly subscribers are engaged for a year, so they use the service better. (Pauline · Pauline (Oct 2024) ▶ 08:38)
• Focus on higher-paying customers — They prioritised higher-paying businesses because they saw lower churn, accepting fewer (38) paying customers. (unnamed guest · Josef Buettgen & Timo Nikolai (Sep 2025) ▶ 10:01)
• Treat complaints as a stickiness signal — A user who takes time to complain is committed to using the software and wants it fixed. (Tibo · Tibo (Dec 2025) ▶ 11:33)

Stories & examples

• Curious sign-ups retain worse: Segmenting by sign-up reason showed people who signed up out of curiosity had lower retention and customer love than those who needed marketing now. Lesson: Sign-up intent predicts retention; track it. (Gaurav · Gaurav (Aug 2026) ▶ 08:38)
• Pray Screen retention: Pray Screen blocked users’ apps until they prayed, giving 60% day-30 retention, helped by the religious pull; it got 30–50 cent installs in the US via Meta ads. Lesson: A mechanic that forces use plus an emotional pull can drive exceptional retention. (Lotts Ezeike · Lotts Ezeike (Nov 2025) ▶ 09:22)
• Stuck at 1-2K MRR despite distribution: Influencer marketing brought steady downloads and trials, but revenue stalled at 1-2K MRR; at the Super Bowl many converted to trial and nobody stayed. Lesson: Strong top-of-funnel with poor retention signals a product problem. (unnamed guest · Eyal & Yahli (Nov 2025) ▶ 03:33)
• A user who saw her cat again: A user whose cat had died found a similar-looking cat in the app and felt she could see her cat again. Lesson: Emotional design creates strong user attachment. (Ryan · Ryan (Jul 2026) ▶ 09:39)

Numbers

• Monthly churn, before and now (target below 5%): 11 -> close to 7 % per month (self-reported, approx.) (Dmytro Krasun · Dmytro Krasun (Nov 2024) ▶ 08:16)
• Churn after going agentic vs before: 13.7% vs 26-28% % churn (self-reported) (Nevo · Nevo (Jul 2026) ▶ 02:50)
• Pray Screen day-30 retention: 60 percent (self-reported) (Lotts Ezeike · Lotts Ezeike (Nov 2025) ▶ 09:22)
• Day-30 free-to-paid retention: 61 % (self-reported) (Presh · Presh (Dec 2025) ▶ 01:57)
• Churn: 31 percent (self-reported) (Roman · Roman (Sep 2026) ▶ 02:20)
• Churn (described as very high, being worked on): 19 percent (self-reported) (Nevo · Nevo (Jan 2026) ▶ 03:00)
• Churn rate: <10 percent (self-reported, approx.) (Miquel Palet · Miquel Palet (Mar 2026) ▶ 02:36)
• Returning customer rate: 40 % (self-reported, approx.) (Jeremy · Jeremy (Mar 2024) ▶ 08:04)
• Retention lift from one commitment screen at the end of onboarding: 3 % retention (self-reported) (Mauro Botanes · Mauro Botanes (Sep 2026) ▶ 12:30)
• Target client retention at product-market fit (vs 3-5 months industry): 10 months (self-reported) (Alex Lieberman · Alex Lieberman (Nov 2023) ▶ 11:00)
• Worked example: 1 new customer/day, 10% churn, $50 each: 200 customers; 10000 MRR customers; USD per month (self-reported, approx.) (Nic · Nic (Feb 2026) ▶ 12:40)

Caveats & warnings

• Cancellation survey reasons alone didn’t help; answers were too generic. (Dmytro Krasun · Dmytro Krasun (Nov 2024) ▶ 07:17)
• Switching to subscription increased churn, though revenue became more predictable. (Fernando Pessagno · Fernando Pessagno (Dec 2024) ▶ 08:53)
• Churn roughly equals new installs so far; marketing hasn’t been hard-launched. (Jack · Jack (Jun 2026) ▶ 02:49)
• Which metrics matter depends on the product; activation is key for Symmetry but most apps should watch first-week and 30-day retention. (Mauro Botanes · Mauro Botanes (Sep 2026) ▶ 09:05)
• A one-off product like Rummer needs constant new-customer marketing or the business dies off. (Jure Sotosek · Jure Sotosek (Apr 2025) ▶ 13:38)
• Tools used by IT admins inside companies probably have very low churn (host’s assumption; numbers not shown). (presenter · Thomas Mahony (Mar 2026) ▶ 11:16)
• Hiro’s churn is the best Pat has seen from any Starter Story guest. (presenter · Zac (Jun 2026) ▶ 02:16)

How much to trust this

• Evidence: 51 episodes, 49 independent speakers (repeat guests counted once; Pat and Gus counted as one voice).