How Encodr picks your first deck

New account, empty library - the recommendations you see first come from a short onboarding step where you pick a few interests and a handful of starter books. But that's the last time your stated preferences get the final say. From there, what Encodr recommends is driven mostly by what you actually do.

Behavior outweighs what you said once

The Library's Recommended row scores every genre by combining a few real signals: a book you finished counts far more than one you started, a book you're partway through counts more than one you never opened, and your onboarding picks count least of all, decaying the further down your original ranking they were. In practice, that means one book you actually finish can outweigh a whole category you ticked as "interested" during signup and never returned to. The row re-ranks itself as you read, not once at onboarding and then frozen.

Why behavior, not stated interest

Stated preference is a weak signal - you tick boxes during onboarding based on what sounds appealing in the abstract, before you've actually spent time with any of it. What you finish is a much stronger signal of what you'll actually keep studying, so weighting it heavier is a deliberate choice, not an oversight. It also means the recommendation gets better the more you use the app, rather than staying anchored to a five-second decision you made on day one.

Cold start still works

If you're brand new and haven't finished or started anything yet, the row doesn't sit empty waiting for behavior data - it falls back to a sensible default category order so there's always something to open. Onboarding's starter-pick step (choosing a few books up front from your interests) exists specifically to seed that first session with something relevant before any real behavior signal exists.

This only applies to the Library

This recommendation logic is scoped to book and course recommendations in the Library - it's separate from the feed's own logic for which class gets studied first when you open the app, which prioritizes what's due soonest or most recently studied. See inside the feed for how that default selection works, and the library for what's actually sitting on the shelves this logic is choosing from.