How do apps keep users coming back? What founders said
By gr0w Research · Updated · From 185 founder interviews · 4 min read
Short answer
In an analysis of 185 founder interviews, the most concrete retention numbers came from apps that built one daily trigger and backed it with streak-loss notifications, widgets and social pressure. Flame's founder said day-30 retention went from 0.8% to 4.2% to about 30% and then 50% over roughly eight months using a daily question, "carrot and stick" notifications and widgets [1]. Others added leaderboards, friend groups, commitment screens and content.
- 0.8% → 50%
- Flame's day-30 retention across its versions, as the founder reported it
- +3%
- retention Symmetry said it gained from one onboarding commitment screen
- 60% vs 20%
- share of Lobby users with 5 friends on day one in Israel versus elsewhere
- 50% = amazing
- year-one subscription retention, per an app acquirer's rule of thumb (40 good, 30 okay, 20 bad)
What is the single biggest retention lever founders described?
A daily habit with a trigger. Anup, founder of Flame (a couples app), said his first version had 0.8% day-30 retention [1]. After he studied habit formation, he concluded that every retention event needed a trigger and a daily action that called people back [1]. Flame went from 4.2% to about 30% and then 50% day-30 retention, which he said beat TikTok's roughly 40% [1].
A key step was finding the feature people actually returned for. That was the daily question, not the AI coach, so he pulled it to the front of the home screen [1]. He also hid each partner's answer until you answered yours, so each person's answer pulled the other back [1]. The whole climb took about eight months of iteration.
How do streaks and notifications work together?
Founders paired a reward nudge with a loss warning. Flame sent a morning "come answer your question" notification (the carrot). If you hadn't answered later in the day, it warned that you would lose your streak and your partner would be sad (the stick) [1]. Anup said the streak mattered because it was tied to the partner, not to the streak itself [1].
Other examples:
- Cal AI: Daniel said the team used streaks so people log meals daily, and borrowed from Duolingo and Strava [2].
- Wrestle AI: George copied Duolingo-style notifications tied to user actions, such as "you didn't track your calories" [3]. He also sends win-back copy to people who fell off, but said he doesn't track which notification gets clicks [3].
Do home-screen widgets help retention?
Flame's founder said widgets were something he discovered too late. On mobile, a widget on the home screen calls the user back every time, and it stands out because almost every app now sends notifications [1]. He pointed to a distance-between-partners widget, noting that widgets are bigger than other icons and can update dynamically [1]. He said Couplejoy was one of the first to put a widget on the lock screen and built its social strategy around it, and Flame copied the idea [1]. The library gives no separate before-and-after number for widgets alone.
Do leaderboards and social features keep people in the app?
Founders said yes, when friends are present. Cal AI added groups where friends log food together and see who has the longest streak, first with friend groups and later with public groups [2]. Symmetry's gym app has ranks per muscle group and a social feed, and the founder noted you "cannot lose your streak to your friends" [4]. Runafy's founder ran a pre-order so the leaderboard filled past 1,000 entries within an hour of launch, avoiding an empty board [5]. Evan built Locked with XP, badges, characters and leaderboards to stand out in a crowded habit category [6].
Roger Chen of Lobby said retention doubled where users had friends: 60% of new users in Israel had 5 friends on day one versus 20% elsewhere. He said content marketing can't fix that density [7][7].
Can onboarding and content change retention?
Yes, in small, testable steps. Mauro of Symmetry added a screen at the end of onboarding where users "sign" a pledge to work out. He said that single screen raised retention 3% [4]. His team A/B tests changes for at least two weeks against a control [4]. He also noted that two very different redesigns of one page moved no metric [4].
George said real retention comes from content inside the app. For Brawler he paid a coach to record instructionals, where a vision model checks your boxing stance before the next lesson. He added a social feed to Wrestle AI [3]. The PushScroll founders said they were working to improve retention within their seven-day free trial so users see value before it ends [8].
What retention numbers count as good?
It depends on the category. Josh at Blue Throne, an app acquirer, gave a rule of thumb for year-one subscription retention: 50% is amazing, 40 good, 30 okay, 20 bad. He said not to apply it to dating apps [9]. He also called resubscriber rates a strong stickiness signal and said day-7 and day-30 retention matter, especially with a free tier [9].
Tibo, who runs several SaaS products, said not to push acquisition before retention is solid, or 99% of users churn right after trying it [10]. Raphael Kramer said his meme game charted on the App Store, but retention was bad [11]. All figures are self-reported.
| Tactic | Founder / app | Reported result |
|---|---|---|
| Daily question + carrot/stick notifications | Flame (Anup) | Day-30 retention 4.2% to ~30%, then 50% [1] |
| Streaks and friend groups with leaderboards | Cal AI (Daniel) | Groups feature built to keep users logging daily [2] |
| Lock-screen / home-screen widget | Flame (Anup) | Called a strong reminder; no separate number [1] |
| Commitment screen in onboarding | Symmetry (Mauro) | +3% retention [4] |
| Friend density on day one | Lobby (Roger Chen) | 60% vs 20% of users with 5 friends; double retention [7] |
| Duolingo-style action notifications and win-back | Wrestle AI (George) | No stats tracked per notification [3] |
| In-app instructional content and social feed | Brawler / Wrestle AI (George) | Founder's view of what drives real retention [3] |
Frequently asked questions
What is a good day-30 retention rate for an app?
The library has no universal benchmark. Flame's founder reported 50% day-30 retention and said TikTok sits around 40% [1]. Blue Throne's rule of thumb is for year-one subscription retention: 50% amazing, 40 good, 30 okay, 20 bad [9].
Should I work on retention before acquisition?
Several founders said yes. Tibo said going broad too early means 99% churn right after users try the product [10]. George said apps that ignore retention are hard to exit [3].
How long did it take to improve retention?
Flame's founder said about eight months of iteration took day-30 retention from 4.2% to 50%, after an earlier version sat at 0.8%. He also reported that investors passed on the 50% retention because there was no distribution plan [1].
Do simple apps need gamification?
Not always. Matt of Post Bridge said the biggest driver was a product so useful that people need it in everyday life [12]. Gamification mattered most in habit, fitness and relationship apps.
Is there evidence on notification copy testing?
Little. George said he didn't track which notification gets clicks [3]. The measured result in the library is Flame's carrot/stick set-up, reported alongside its other changes.
Sources
Founder interviews on YouTube; numbers are as the founders said them. How this research is done.
- [1]Full App Growth Guide 2026 - How to Make Viral Content to Get Users & Make Your App Retentive · Flame (Superwall)
- [2]I speedran a new app from 0 to $2m/year in 6 months (here's how) · Sway AI, Cal AI (Superwall)
- [3]Meet the ex-wrestler making $41k/mo with AI apps · Wrestle AI, Brawler, AbMax (Superwall)
- [4]How I Built This $160K/Month Gym App · Symmetry (Starter Story)
- [5]I speedran an ai app from 0 to $100k exit in 26 days · Runafy (Superwall)
- [6]I'm 14 And I Built A $14K/Month App · Locked (Starter Story)
- [7]How I Built the #1 App on The App Store (Twice) · Lobby, Bro (Superwall)
- [8]Our App Makes $30K/Month Profit Using This Simple Strategy (copy us) · PushScroll (Superwall)
- [9]$172M of App Building Advice in 50 Minutes · Blue Throne (Superwall)
- [10]I Built 4 SaaS Apps to $100K MRR: Here's My Exact Playbook · Revid.ai, Outrank, SuperX, PostSyncer, Feather (Starter Story)
- [11]this 14 yr old makes $100k/mo PROFIT building apps with ai (here's how) · The Baby game, Skibidi Toilet mobile game, Debloat AI, Amanda AI (Superwall)
- [12]I Built a $10K/Month App from My Mom’s Basement · Curiosity Quench, Post Bridge (Starter Story)