I Tracked 28,000 People Trying to Build a Daily Habit - Heres When They Quit

A solo developer tracked 28,547 users of the AI-powered habit-building app Wishyze and discovered a four-phase retention model: Spark (days 1–7), Void (weeks 2–6, where 73% quit), Alignment (weeks 6–12), and Manifestation (12+ weeks). By showing users what week 8 looks like before they reach it, the developer improved Void retention by 22%.

Six months into building Wishyze, I was ready to give up. I'd spent months building what I thought was a solid app: AI-powered daily rituals, affirmations, visualizations — the works. People signed up. They even used it. For about four days. Then they vanished. I stared at the Supabase dashboard one night, watching the retention curve plummet like a ski slope. Day 1: 100% active. Day 3: 40%. Day 7: 12%. By day 14, barely 4% of users were still showing up. I had built something people clearly wanted to use — they signed up — but couldn't keep using. That gap nearly broke me. When you're a solo founder, you don't have a data science team. You have a SELECT query and existential dread. So I did what any desperate developer does: I wrote a script. -- I lived in this query for weeks SELECT user id, DATE created at as ritual date, COUNT OVER PARTITION BY user id ORDER BY created at as ritual number, DATEDIFF created at, LAG created at OVER PARTITION BY user id ORDER BY created at as days since last FROM rituals ORDER BY user id, created at; What emerged changed everything. I mapped 28,547 users across their journey and found a pattern so consistent it was almost eerie. Every user who stuck around went through the same four phases. Every user who quit? They all hit the same wall. Phase 1: Spark Days 1–7 This is the honeymoon. You've just discovered the thing. Everything feels fresh. The AI generates a ritual that feels personalized . You're excited. Retention is great. Everyone's a believer. Phase 2: Void Weeks 2–6 This is where 73% of users quit. Not 20%. Not half. Seventy-three percent. The Void is brutal because nothing is "wrong." The app still works. The rituals are still good. But the novelty is gone. You wake up and think, "Do I really want to open that app today?" And the answer, for most people, is no. I lost sleep over this phase. I tried notifications annoying . Streak counters didn't help . Gamification felt cheap . Nothing moved the needle. Then a user emailed me — not a feature request, just a journal entry: "I almost quit last week. Then I saw I'd done 14 days in a row and realized I actually felt different. Less anxious. More intentional. I didn't notice the change happening — I only noticed it when I almost stopped." That email gave me the answer: users can't feel the compound effect while it's happening. They quit right before the shift, because the shift is invisible from the inside. So instead of nagging users to stay, I started showing them what week 8 looks like before they reached it. Little previews. Testimonials from people who'd made it through. A "you're in the Void — this is normal" message at day 10. Not a notification. A recognition. That one change improved Void retention by 22%. Phase 3: Alignment Weeks 6–12 If you survive the Void, something shifts. The ritual stops feeling like a chore and starts feeling like a reflex. Users describe it as "I don't think about it anymore — I just do it." This is where the magic happens. The affirmations stop sounding silly. The visualizations start feeling real. The skeptic becomes the advocate. Phase 4: Manifestation 12+ weeks I won't pretend everyone reaches this. But the ones who do? They're the reason I keep building. The longest streak on Wishyze is 93 days. That user started a side project, left a toxic job, and — her words — "stopped waiting for permission to live my life." She's not a power user because of badges or points. She's a power user because the daily ritual became her thing, not my app's thing. That's the real goal. The Phase Model didn't just become a naming convention. It rewired how I built the app: The ritual adapts across phases. Spark rituals are quick wins. Void rituals deliberately surface the struggle. Alignment rituals build complexity. The AI prompt knows the user's phase. When DeepSeek V4 Pro generates an affirmation, it includes context: "This user is in Void phase, day 18. They've missed 3 of the last 5 days." The output changes — more compassionate, more specific, less "you can do it" cheerleading. I stopped optimizing for signups and started optimizing for week 3. Every feature decision now has one question: will this help someone on day 19 show up on day 20? Nobody tells you that the hardest part of building a habit app isn't the AI, the auth, or the payment integrations. It's not even the churn. It's facing the possibility that what you're building might not work — and having no one to blame but yourself. As a solo founder, there's no scrum master to unblock you. No PM to blame for bad requirements. When retention tanks, it's because you didn't understand your users well enough. That's lonely. It's also clarifying. I learned to stop guessing and start measuring. Every feature now ships with a hypothesis, a metric, and a deadline. If the metric doesn't move in two weeks, I rip it out. No ego. No sunk cost. Just data and discipline. The Phase Model is far from solved. I'm currently experimenting with Phase-aware onboarding — basically, telling new users on day 1 what Void will feel like, so it doesn't catch them off guard. Early results are promising: pre-warned users are 30% more likely to reach day 14. I'm also exploring what happens when you bring social accountability into the Void phase — not leaderboards or pointless competition, but small, private accountability pairs. Two strangers, same phase, one daily check-in. No chat. No profile. Just "did you do it today?" If that sounds interesting, or if you're building something similar and want to swap notes, I'd genuinely love to hear from you. The best thing about building in public is that you don't actually have to build alone. If you're curious about the Phase Model or want to try a daily ritual yourself, you can explore it at wishyze.com — the free tier gives you 3 rituals a day, no card required. I built it for people like us: the ones who know they want something more, but aren't quite sure how to get there yet.