The app was always built to show what repeats, in what order, and how far apart. What the site did not say, until today, is why anyone would want that. The answer is a sentence Timothée Chalamet said in an interview: you have to realise life is coming from you and not at you, and that takes time.
That is the whole idea. Anyone can say what went wrong last year. Saying what you were doing in the weeks before, how often, and in what order, is harder, and it is the part that was yours. A log kept for a few weeks can show that part, with a count. Then the reading is the person's own: the app keeps a place under every finding for what you think is behind it, in your words, and never writes one for you.
What changed today: the home page says this out loud, the answer pages carry one line making clear the log is yours, and the free Daylio statistics page gained the same box under every finding. The app gets the box too, next to the "not now" and "this isn't right" buttons that were already there.
What did not change: the app still never says why. It reports what lines up, what you kept doing in the days before things went wrong, and keeps a place under each finding for what you make of it.
Three tools for a Daylio backup
Three pages went up today under /tools. Each one reads a .daylio backup file in your browser and hands something back: the same CSV Daylio exports plus a JSON file, one Markdown note per day for an Obsidian vault, or the statistics this app runs, run over that log. Nothing is uploaded. The one thing a page sends out is a count of how many files were converted and whether it worked.
The backup is the file to build on. Daylio exports a CSV on its own, free, but the CSV writes each mood as a name only. The backup keeps the mood's place on Daylio's five-step scale, plus the tag groups and favourites, and it is the file people still have once the app is gone: the one on Google Drive or iCloud, the one moved between phones.
The parser is open source. It is on GitHub as daylio-backup, MIT licensed, with no dependency beyond the zip reader. If you write code, you can use it without touching these pages.
On a Daylio log, the tests that need energy or sleep are left out, and the ones that remain are the ones a mood-and-activities log can support. The statistics page says so, on the page. Apple Health is the export to look at next. It carries sleep, a second number per day, which the lagged correlation test needs.
The site was on the wrong shelf
This site went up on 20 August with eight pages of answers. Six of them were written as though the app were a mood tracker: how to track your mood, what to log, how to keep the habit going. That framing was wrong, and framing is what decides who ever finds a site like this.
A mood tracker answers how it has been. It charts the month and leaves the reading to you. This app is built for a different question, the one that turns up at the end of a bad week: why does this keep happening. It can't answer the why, and it never will. What it can do is show what repeats, in what order, and how many days apart the two ends sit.
So the answer pages have been re-cut around that. There are still eight of them, and 79 questions where there were 71. Six of the eight are new. They cover why this keeps happening, what comes before a bad stretch, finding what sets you off, when a good stretch ends, how the app decides that something is a pattern, and why this is not a mood tracker. The two that were already right, on paying once and on the journal staying on your phone, are unchanged.
Two things also got added that were missing. Every answer page now carries the date it was last edited, which is a small thing that no page on this site had. And the page about how the statistics work now cites where those methods come from, with links, so nothing on it has to be taken on trust.
The old addresses still work. They point at whichever new page carries the closest thing to what they used to say.