WriteDaily at Year Two: Lessons from 25,000 Writing Sessions

WriteDaily has been running for two years now. What started as a side project — a 750-words-a-day writing tool with LIWC sentiment analysis — has quietly processed over 25,000 writing sessions from a small but loyal user base. (When I wrote the year-one recap, we’d just crossed 10,000 sessions.)

I’ve been maintaining it alongside client work, and the slow, steady iteration has revealed lessons that only come from running something for years rather than months.

What Two Years of Data Reveals

The Habit Curve Is Real

Users who write for 7 consecutive days have a 68% chance of still writing 30 days later. That number jumps to 84% for users who reach 21 consecutive days. The habit formation literature claims it takes 21–66 days to form a habit. Our data aligns: 21 days is a real inflection point.

Morning Writers Are Different

Sessions started before 10am average 812 words and have an 84% completion rate. Sessions started after 8pm average 623 words with a 51% completion rate. The data is unambiguous: morning writing is both more productive and more consistent.

Sentiment Patterns Track Life Events

I’ve had users reach out to say their LIWC charts reflected major life changes before they consciously acknowledged them. One user noticed her “positive emotion” score dropped from 78% to 31% over a two-week period — she was diagnosed with depression shortly after. Another saw his “achievement concern” spike during a job search he hadn’t yet told friends about.

Writing analytics, it turns out, is also self-awareness analytics.

What’s Working Well

The core loop continues to deliver. The editor auto-saves reliably (zero data loss in two years across 25,000 sessions). The pace-tracking graph gives writers real-time feedback on their flow state. The streak calendar — simple green squares for completed days — is still the most effective motivator in the product.

The LIWC parser has taken on a life of its own. I open-sourced it, and it’s been used in a linguistics research paper and a Python port that someone else maintains. That’s more impact than I expected from a weekend’s worth of PHP.

What’s Evolving

Not everything scales. The original PHP codebase — built fast with raw queries and inline JavaScript — is becoming a bottleneck. Each new feature requires touching files that should have been refactored months ago. The technical debt is manageable but growing.

I’ve been working on a framework migration in parallel — porting the domain logic to Laravel with proper Eloquent models, migrations, and a test suite. The goal isn’t a rewrite from scratch; it’s an incremental migration that keeps the existing product running while the backend gets healthier.

The sentiment analysis pipeline is also getting an upgrade. Currently, LIWC runs fresh on every page load. For a growing user base, that’s inefficient. The migration adds cached sentiment arrays to the entry model, so analysis runs once on save and subsequent views pull from cache.

What I’m Thinking About Next

1. Gamification. I’ve been designing a badge and condition system — inspired by the gamification framework from PrestaShop, where I’ve been doing client work. The idea: users earn badges for hitting milestones (first 750-word entry, 7-day streak, 50K total words). The architecture uses declarative JSON conditions evaluated by a cron-driven engine. It’s solid on paper; implementation is in progress.

2. Authentication. WriteDaily was originally single-user. Adding Sentry for proper multi-user auth opens the door to a broader audience.

3. Pricing. The tool has always been free. I’m considering a modest subscription — enough to justify continued development time but not so much that it excludes the people who benefit most.

4. Community features. Several users have asked for shared writing challenges and group word-count goals. Leaderboards and community streaks could turn a solitary practice into a social one.

The Reality of Long-Running Side Projects

WriteDaily hasn’t had a dramatic trajectory. It grew slowly, plateaued, and settled into a steady state. It’s never been a startup. It’s a tool that a few hundred people use regularly, and that I use daily.

The maintenance burden is real. Client work always takes priority, so WriteDaily gets attention in the gaps. But the gaps add up. Two years of evenings and weekends have produced something I’m genuinely proud of — a habit-forming writing tool with data-driven insights that nothing else offers.

If you write regularly and want to understand your own patterns, try it. The data might surprise you.