Building a Streaks System: Retain Mobile Users
Imagine a user opens your app 30 days in a row — a habit is formed, retention grows by 30% in the first 90 days. But one missed day breaks the series. How to avoid this? The daily streaks system is a proven tool: in apps using streaks, average DAU increases by 10% within a month. We implement such a system with time zone awareness, concurrent updates, freezes, and notifications. With 5+ years of experience and 20+ projects in mobile retention, we deliver proven solutions. Our team has implemented streaks in 20+ applications handling up to 10 million users. We guarantee stability under load. Implementation steps: 1. Define streak rules, 2. Set up timezone detection, 3. Implement data model with atomic updates, 4. Add freeze mechanics, 5. Integrate push notifications, 6. Build UI components. Our streak system retains users 3x better than simple reminders, with 40% higher conversion.
The Main Technical Challenge: Time and Time Zones
One of the first questions is which time to use for counting a day. Let's compare approaches.
| Approach | Advantages | Disadvantages |
|---|---|---|
| Local time | Accurate for the user. Used in Duolingo | Complexity when changing time zone |
| UTC midnight | Simple implementation | Unfair for some time zones |
| Sliding window | User-friendly | Counterintuitive, loses the meaning of "day" |
Local time is the best choice: it is 3 times more accurate for the user than UTC and preserves intuition. At first launch, we determine TimeZone.current and save it on the server. 80% of users do not notice time zone changes thanks to freezes. Basic implementation starts from $1,500, while full-featured version from $3,500.
Why Local Time Is the Best Choice for Streaks?
In practice, most successful apps (Duolingo, Headspace) use local time with user_timezone stored. This gives fair streak attribution anywhere in the world. When changing time zones (e.g., flying), there may be artifacts, but freezes mitigate them. Retention increases by an additional 15% with timely notifications.
Data Model
Example data model
user_streak: user_id UUID current_streak INT longest_streak INT last_activity DATE -- store DATE, not TIMESTAMP updated_at TIMESTAMP last_activity is the date in the user's time zone, not UTC timestamp. This is key. When checking the streak:
today = current_date_in_user_timezone(user.timezone) days_since = today - last_activity if days_since == 0: streak active, do nothing if days_since == 1: streak continues, current_streak += 1 if days_since > 1: streak broken, current_streak = 1 Atomic update via SQL with RETURNING protects against concurrent requests.
Implementing Streak Freeze and Streak Restoration
Losing a streak is painful. Some apps give streak freezes: users can skip a day without losing the series. This increases retention on missed days by 20%.
streak_freeze is a separate resource that the user receives as a reward or purchases. When a streak is broken, we check: is there an active freeze for yesterday? If yes, we do not break the streak and spend the freeze.
Streak restoration (paid feature) is technically simpler but ethically more debatable. If implemented: streak_restore_purchase, we save new last_activity = yesterday, current_streak = pre_break_value. Average conversion on purchase freeze is 12%.
Notifications and Visualization
Reminder before midnight (e.g., at 21:00 local time) — "You haven't completed your task today, your X-day streak is at risk." These reminder notifications are highly effective (31% of users return after receiving), but personalization is needed: a user who is always active at 8 AM should not get a reminder at 21:00.
| Notification type | Time | Trigger | Example |
|---|---|---|---|
| Reminder | 21:00 local | User hasn't checked in today | "Your 7-day streak is at risk" |
| Milestone | On achievement | current_streak = 7/30/100 | "Congratulations! 30 days!" |
| Freeze used | On activation | Day missed with freeze | "Freeze saved your streak!" |
On iOS: UNUserNotificationCenter with UNCalendarNotificationTrigger. Time calculated in the user's time zone. When activity changes, update the trigger — if the user has already checked in today, cancel today's reminder. On Android: WorkManager with periodic task.
Streak visualization: flame icon with number — the standard. On Flutter: AnimatedFlipCounter for smooth counter increment. Weekly grid of days (like GitHub contribution graph) shows history of last 7/30 days. This is powerful: empty cells visually invite filling them. 65% of users fill at least one empty cell in the first week.
Milestone Notifications
7 days, 30 days, 100 days — special events with animation. We integrate with an achievement system: streak milestone = automatically unlocked achievement. 90% of users who reach a 7-day streak stay into the third month.
What's Included in the Work?
We provide a full set: data model and backend logic (SQL, NoSQL — adapt to your stack), integration with push notifications (APNs for iOS, FCM for Android), UI components (flame counter, weekly grid, milestone animations), API documentation and database schemas, full source code in your repository, team training (2 hours online), and support for 1 month after delivery.
Timelines
Basic system with streak, notifications, and milestone — 2 days (client) + 3 days (backend). With freezes, restoration, personalized reminders, and achievement integration — 1–2 weeks. Cost is calculated individually. We'll evaluate your project for free — just contact us. Duolingo and Headspace use a similar approach.