Streaks are one of the most effective habit tools ever invented, and one of the easiest to get wrong. A streak that helps you on day 10 can start hurting you on day 80. You stop coding because you want to and start coding because you're scared of a number going back to zero.
I build a streak-based app, so I think about this a lot. Here's what I've learned about keeping a streak useful and stopping it turning into a source of anxiety.
Why streaks work, and why they go wrong
A streak works because it changes the question you ask yourself. Instead of "should I code today?" (an open negotiation), it's "do I want to break the streak?" (an easy no). That's powerful on low-motivation days.
It goes wrong in three ways:
- The floor is too high. If keeping the streak costs an hour, a long streak becomes an hour of mandatory work every day, forever. That's not a habit. It's a second job.
- Missing is catastrophic. If one missed day takes a 90-day streak back to zero, the streak's value grows every day, and so does the fear of losing it. Eventually the fear is bigger than the benefit.
- The streak replaces the goal. You start doing the minimum to keep the number alive, and stop caring about the project the number was supposed to serve.
Each one has a fix.
Fix 1: a floor you can hit while sick
Your daily minimum should be something you could do with a cold, on a travel day, or in exam week. For coding, that's one small, real commit. Not an hour; not a feature.
A small floor sounds like it lowers the bar. In practice it raises your output, because the habit survives the bad weeks. Most good days will go far past the floor without any pressure.
Fix 2: a missed day is repairable, not fatal
The most important design choice in any streak system is what happens when you miss.
"Back to zero" is simple and brutal. It makes the streak fragile in exactly the way that causes burnout. A better model:
- A miss is visible. You should know you missed. Pretending it didn't happen makes the streak meaningless.
- A miss is repairable. You get a limited way to repair occasional misses, so one bad day doesn't erase months.
- A repair is honest. A repaired day is marked as repaired, not passed off as a normal day.
That's how we built Vigilante. If you miss a day, Revival Fluid can repair it. The repaired day is marked as mended. It never counts as a verified day and never earns that day's reward, so your record stays honest. Fluid is limited on purpose:
- free players can earn one a month through play (finishing a 7-day check-in calendar or mission stages);
- Pro members get two a month;
- you can also buy packs.
It's meant for the bad day, not for skipping every Sunday.
Fix 3: missing shouldn't destroy anything
When you miss a day in Vigilante, your town doesn't burn down. The salvage that builds up while you keep your commitment stops, and starts again when you come back. Nothing you've built is taken away.
That was deliberate. A system that punishes you for missing teaches you to dread opening it, and an app you dread is an app you delete. We even have a rule for our own emails: they never claim progress is lost, and they never use fear or shame.
If you're designing your own tracking system, even a spreadsheet, apply the same rule: a gap should pause progress, not reverse it.
What about rest days?
Honestly: Vigilante doesn't have scheduled rest days today. Your goal is daily. We've gone back and forth on this, and here's the thinking.
The case for rest days: humans need rest, and a daily rule can feel relentless.
The case against: for a *coding habit*, the cost of a full day off is mostly re-entry friction. A two-day weekend gap is enough to lose the thread of a project. A 15-minute floor commit on Saturday is less effort than spending Monday working out where you were.
What we've landed on: keep the daily floor tiny, so the "rest" is built into the size of the floor rather than into days off. On a day you'd call a rest day, you push one small thing and you're done. And on a day you truly can't, there's a repair.
There's also a daily check-in in the game that works even on a day with no commits. It gives a small reward for showing up, without counting as a verified day. It keeps you connected to the habit without pretending you coded.
If your life needs real days off, like a weekly religious day, a caregiving day or a fixed family day, a strictly daily target may not suit you, and that's a real limitation of our design.
Fix 4: keep the project bigger than the number
The surest sign a streak has gone bad is when you'd rather keep the number than do good work. Some checks:
- Look at what you built, not the count. Once a week, look at what's actually different in your project.
- Let good days be good. The floor is a minimum, not a target. If you're in flow, keep going.
- Let bad days be small. If you hit the floor with something tiny but real, that day was a success. Don't add guilt.
- End streaks on purpose if you need to. If you finish a project, or a season of life changes, you can stop. A streak you end by choice isn't a failure.
Warning signs of streak burnout
- You feel relief, not satisfaction, when you hit your target.
- You've started making trivial commits just to keep it alive, every day.
- You check the streak more than you think about the project.
- A missed day ruins your mood for longer than the day itself.
If you see two or more, lower your floor, plan a repair, or take a break on purpose.
The healthy version
A healthy streak is small, honest and forgiving:
- small enough to keep on your worst day;
- honest about misses and repairs;
- forgiving enough that one bad day costs a day, not a season.
That's what I try to build, and it's what I'd look for in any habit tool.