Home / Improve my conviviality score
Improve my conviviality score
A field guide for people who make software. It shows what each of the six questions looks at and which changes move a score most.

Download the field guide
A free 12-page PDF for people who make software: twelve changes that move the score, what reviewers look at for each question, a plain answer on money, and a worksheet.
Download the PDF PDF, 12 pages Read the summaryStart with the quick list
These twelve changes are roughly in order of effort against payoff. The first five usually take an afternoon each, and the last few need a decision from you.
- Add a full export in an open format (JSON, CSV or Markdown). (Lock-in)
- Remove analytics and third-party trackers, or make them opt-in. (Power)
- Remove streaks, nag screens and fake urgency. (Self-direction)
- Write in plain words what happens to people's data. (Power)
- Add a Limitations section to the README, and keep marketing copy true. (Delivery)
- Drop dependencies and background work the job doesn't need. (Footprint)
- Let people use it without an account unless the job really needs one. (Lock-in)
- Store data on the person's own device wherever no server is needed. (Longevity)
- Make the core job work offline. (Footprint)
- Document how to run it yourself, or say plainly why you can't. (Longevity)
- Publish the source under an open licence and link to it. (Power)
- Keep export and read access working after someone stops paying. (Lock-in)
Then work through the six questions
The PDF gives each question a page with what a reviewer looks at, what a 1, a 3 and a 5 look like, what to change this week and this month, and the evidence worth showing. The questions are the same ones used on How we score.
- Footprint. Does it run light, work offline, and keep going on an old laptop, or does it need new hardware and a data centre?
- Lock-in. Can you leave with your stuff in an open format, or does it make you depend on a whole category of product to take part?
- Self-direction. Can a normal person learn it by using it and point it at their own purposes, without dark patterns training their habits?
- Power. Who holds the controls? Team size, governance, open source, whether the price shuts people out, and whether your data is the product.
- Longevity. Does it keep working if the company folds or you stop paying, and does it respect the hardware you already own?
- Delivery. Does it actually do the job, without ads, engagement bait, or a promise the product can't keep?
What about money?
Conviviality scores how a tool earns, and it doesn't score whether the tool charges. Cryptomator charges about 30 euros for mobile editing and scores 27, and Bitwarden has paid plans and scores 25. One-time purchases, subscriptions for hosting or sync, paid support, and donations don't cost a tool any points. Advertising, selling data, charging to export and trial traps do. We have no evidence that a higher score earns more, and the PDF says so, along with a way to count what you would give up. The short version: charge for the service, not for the exit.
Using an AI coding agent?
The same audit exists as a prompt you can paste into Claude Code, Codex or Cursor. See For AI agents.
When you're ready
Use the suggestion form to submit or re-submit your tool. It prepares a message for you to check and send. We review independently, date the result, and a score can change as the tool does.