Why Your Side Projects Die After 3 Weeks (And How to Keep Them Alive)

Have you abandoned a side project this year?

You are not alone. I have a folder on my laptop called archive/ that I do not open anymore. Inside it are seven repositories. Each one started with a clean README, a working dev environment, and a real plan. Each one stopped getting commits sometime around week three.

If that sounds familiar, this is for you. I have been talking with other side-project builders for a while now, and the pattern is almost embarrassingly consistent. Week one is excitement. Week two is momentum. Week three is when it goes quiet. By week four, the project is on the shelf next to all the other ones.

It is not because you ran out of ideas. It is not because the idea was bad. The death of a side project is a structural problem, and once you see the structure, you can do something about it.

What actually happens at week 3

I used to think I was just lazy. That was the easy explanation. Then I started tracking what I did every day, and the data told a different story.

There are three failure modes, and they almost always happen together.

1. Context switching cost. Your side project lives in Notion. Your tasks live in a Trello board you set up the first weekend. Your design references are in browser bookmarks somewhere. Your code is in a folder you have to dig for. By week three, just sitting down to work on the project costs you fifteen minutes of "wait, where was I?" before you write a single line of code. That cost compounds. Eventually it is cheaper to scroll Twitter.

2. Priority drift. A new idea shows up around week three. It always does. The new idea has the shine the current project has lost, because you have not yet hit any of its hard parts. You start telling yourself you will "switch over for a week, then come back." You do not come back. You start a third project. The cycle repeats.

3. Tooling fatigue. You spent the first weekend setting up the perfect stack. New framework, new linter config, new project template, new task manager. By week three, the tooling has not actually helped you ship anything. It has just become more surface area to maintain. You quietly resent it.

The thing none of these failure modes have in common is "I lost interest." Interest is downstream. The structure killed the interest.

The conversation I keep having with myself

The honest version sounds like this:

"Did I work on it today? I do not actually remember. Is the task list current? Probably not. What was I doing last? Let me check the commits. Wait, I have not pushed in five days. Was that intentional or did I just stop?"

If you have had a version of this conversation, you know how draining it is. The problem is not that you lack discipline. The problem is that nothing in your environment is helping you remember why you started, or where you left off, or what is next.

I tried writing a paper journal. I tried a Notion dashboard with rollups. I tried a calendar that I respected for nine days and then ignored. They all failed for the same reason: they were extra steps. They lived somewhere I did not naturally go.

So I changed the approach. Instead of finding the right tool, I asked: what is the single screen I already open ten times a day, no matter what? The answer was obvious. Every time I opened a new tab in Chrome, that was a free, unforced touchpoint with my own attention.

That is the screen that needed to do the work.

The small system that finally kept mine alive

I am going to describe a system rather than sell a tool, but I will be transparent: I build STACKFOLO partly because I needed this system to exist for myself. The principles below are what matter. You can implement them with whatever stack you like.

1. Make the project unforgettable

Your project needs to be the first thing you see when you open a browser tab. Not buried in a Notion sidebar. Not in a folder you have to expand. Visible.

I put my active projects on my new tab page in a grid. Five tiles. Color coded. Each one has a status, a next task, and a one-click "open everything" button that loads the local dev URL, the GitHub repo, and the design doc into a tab group. The cost of resuming a project dropped from fifteen minutes to about three seconds.

If you cannot put it on your new tab, put it as your wallpaper. Put it as your terminal MOTD. Put it somewhere it cannot hide.

2. Capture today before you forget today

Every project I have killed had the same forensic evidence: I stopped logging what I did. Not what I planned. What I did.

I now do a two-minute daily log at the end of each work session. Mood, one-line summary, what I shipped, what blocked me. That is it. Here is roughly what it looks like in my system, stored as a small JSON record per day:

{
  "date": "2026-05-14",
  "project": "stackfolo",
  "mood": "focused",
  "summary": "Wired up the daily log AI summary. Refactor still pending.",
  "shipped": ["daily-log endpoint", "telemetry guard"],
  "blocked_by": ["AI prompt eats too many tokens, need to chunk"]
}

The format does not matter. The frequency does. When week three hits and you cannot remember why you cared, you can scroll back and read past-you explaining it. That is the rope that pulls you out.

3. Convert intention into a routine, not a goal

"Ship the MVP" is not a habit. "Work on the side project for ten minutes after my morning coffee" is a habit. The first one is a wish. The second one shows up on a calendar.

I added a single recurring routine block: ten minutes, every weekday, before the workday started. That is it. Ten minutes. The rule is to open the project, look at the daily log, pick the smallest next task, and either do it or write down exactly what is blocking it. Most days I run over and do thirty minutes. Some days I do the minimum and feel fine about it.

This is the move that broke me out of the week-three pattern. The streak does the motivation. You do the showing up.

4. Make progress visible without effort

Open your GitHub profile. The contribution graph is doing real work on your psychology, but only if you are actually committing. I keep a unified timeline of every commit across every account and every project on the same dashboard I use to plan. When the graph is green, the project breathes. When the graph goes flat for two days, I notice on day three, not week three.

You do not need a tool to do this. A weekly cron that emails you your commit count is enough. The point is: instrument the thing before it goes wrong, not after.

What changed for me

I did not finish every project. I am not going to pretend that. But the abandonment rate dropped from "almost everything" to "the ones that genuinely deserved to die." The signal got cleaner. When I quit a project now, I quit it on purpose, with a written reason, and I do not feel the same guilt I used to.

Week three is still the test. But the test is now passable, because the system carries the part of you that would have forgotten.

If you want a single browser-based starting point that ties the new tab dashboard, the daily log, and the routine streak together into one place, try STACKFOLO free on the Chrome Web Store. It is free for the first five projects, which is more than enough to test whether the new-tab-as-hub approach works for you.

And if it does not? Build the four habits anyway. The habits are the thing. The tool is just the rope.