Our first commit went in on September 29. Six days later the site was live. Four days after launch, it still had zero users.
This series is the build log of that site. It starts at the first commit and runs up to whatever we're fighting with this week. Expect the architecture choices, the bugs, the growth experiments and the money, including how little of it there is so far.
What Tintly is
Tintly is a free coloring site. You pick a picture, tap a shape, and it fills with color. There's nothing to install, and you don't need an account to start or to save your progress. Most people use it on a phone.
What it does today:
- About 300 pictures, from easy ones with big shapes to detailed ones with a lot of small areas.
- Tap to fill, or paint with a brush. Undo works, and your progress saves in the browser as you go.
- Every free picture can also be printed as a PDF, on US Letter or A4.
- There's a new picture to color every day.
- No ads.
Under the hood it's a .NET backend with two APIs, a Next.js site, Postgres, and a coloring editor drawn on a canvas. Each of those gets its own post.
Two kinds of posts
The posts alternate between two kinds.
The origin episodes go in order. They cover why we built a website when the first plan was a mobile app, how the backend is laid out, why we swapped our auth provider on the second day, how a flat picture becomes shapes you can tap, and the launch.
The field notes come from the daily work. Usually it's a number that looked wrong and what we did about it, including the fixes that didn't work.
What we promise
The numbers are real and come from our own analytics, with our own visits filtered out. Some of them are small. A few are embarrassing. Over four days in October, 389 people started coloring a picture and 10 finished one. That number gets its own post.
When something went wrong because of a choice we made, we'll say it was our choice. We won't make up anecdotes or quotes. If we don't know why something happened yet, we'll say that too and come back to it when we do.
There will be code where it helps, mostly C# and TypeScript, kept short enough to read on a phone.
Who it's for
Mostly developers and small teams building their own products. If you work in .NET or Next.js you'll recognize a lot of it. If you're running something on a very small budget and wondering whether the boring choices hold up, this is our answer, one post at a time.
How often
A new post every week for the first month, then every two weeks after that. Follow us here to get them as they come out.
Coming up
- We planned an app and built a website
- 389 people started a picture. 10 finished.
- Free pages stay free
- Eight seconds on a phone
- Two front doors: a public API and an admin API
The first one is about the day the plan changed.
Top comments (2)
389 started and 10 finished is the number i'd chase first. do you know yet if people quit because the pictures are too detailed or just because it's a phone and they get bored
Mostly neither. On phones, 64% of sessions end within 30 seconds of the first tap, before detail or boredom can matter. Past that point detail does count: easy pages get finished, detailed ones mostly don't. We also had a Save button that opened a sign-in prompt about 11 seconds in.