The classic ratio is 25/5. The one people actually search for most often alongside it is 50/10 — it shows up attached to three separate searches about focus timers and study music, which is a decent sign it is not a niche preference. A smaller group looks for 90/15, and almost nothing is built for them.
Which ratio suits you is mostly a question of what the work is:
| Ratio | Suits | Breaks down when |
|---|---|---|
| 25/5 | Starting when you do not want to. Tasks that come in small pieces — flashcards, problem sets, admin. | The work needs a run-up. You stop at 25 minutes exactly when you had finally loaded the whole problem into your head. |
| 50/10 | Reading, writing, anything where the first ten minutes are spent getting back to where you were. | You are avoiding the task. Fifty minutes is too big a first commitment on a bad day. |
| 90/15 | Deep single-thread work you already want to do. Long-form writing, proofs, code, revision on a subject you are inside of. | You are interrupted by other people. A ninety-minute block that gets broken at minute forty is worse than two thirty-minute blocks. |
A practical rule: pick the ratio that matches how long it takes you to get back into the work after stopping. If that is thirty seconds, short blocks cost you nothing. If it is ten minutes, a 25/5 rhythm spends a third of your day on re-entry.
What a long block asks of the music
This is the part that gets skipped. A twenty-five-minute block will be covered by one or two tracks, so almost anything works. A ninety-minute block puts the music in the room for an hour and a half, and three properties start to matter that did not matter before:
- No arc. A track that builds is a track that asks for attention at the moment it peaks. Fine once; four times in ninety minutes and it is running your session instead of backing it.
- No change of ensemble mid-piece. Instruments arriving and leaving is an event, and events pull the eye off the page. What you want is a texture that starts and stays.
- Enough of it not to repeat. The obvious one, and the one that becomes an arithmetic problem rather than a taste problem.
Notice these are the opposite of what makes music good to listen to. Music built to be followed is built out of arcs and entrances. Music built to be worked underneath is built out of neither, which is why "put on an album you like" so often fails at ninety minutes and works fine at twenty-five.
The arithmetic, and how to run it on whatever you use
Take the pool of music you would actually play for one kind of session, add up its running time, and divide by the session length. That quotient is how many times through the pool a single session costs you. Under 1.0 and you will hear a repeat inside one sitting.
We keep this as a hard floor per sound, because it is easy to break by accident:
| Session | Pool a sound needs |
|---|---|
| QUICK 25/5 | 30 minutes |
| STANDARD 50/10 | 60 minutes |
| LONG 90/15 | 105 minutes |
Here is what that floor is worth in practice. Cleaning the catalogue, we found three tracks labelled rain that were really renders of an ambient prompt, and had to drop them. That took the rain pool from 106 minutes to 94 — under the 105 that a 90/15 session needs. The obvious repair is to add three tracks back. It does not work: the three replacements added 10.7 minutes against 12.0 lost, so the pool had to be refilled with four, not three. Removing n tracks does not mean adding n tracks; it means adding minutes.
Nobody publishes this number, which is why it is worth running yourself. If you use a playlist, its total running time is right there. Divide.
The other thing we measure, and why the labels are not enough
Our four sounds — lofi, ambient, rain, instrumental — are not separated by what we felt like calling them. They are separated by spectral centroid, roughly "where the energy of the sound sits", from darkest to brightest:
rain < ambient < lofi < instrumental
Every track is measured, and a track whose measurement lands outside its mood's band gets moved or dropped, regardless of what the source file was called. That is how the three mislabelled rain tracks above were caught: not by listening and disagreeing, but because their numbers sat in the wrong band.
One consequence we did not expect. Our loudness-normalisation stage trims rumble below 30 Hz and pulls back the sub-250 Hz energy — which removes low-frequency weight and therefore pushes the measured centroid up, by 113 Hz on average across the whole library. The ordering survives and the gaps hold, but the thresholds calibrated on the unprocessed masters no longer described the files we were shipping. Any figure that survives a processing step without being re-measured is a figure that has quietly stopped being true.
What this means if you use Bloc
Bloc Focus Music ships 25/5, 50/10, 90/15 and a custom length, with 153 tracks across the four sounds. Current pools: lofi 209 minutes, ambient 199, rain 109 — all three clear the 105-minute floor for a 90/15 session.
Instrumental sits at 69 minutes, which clears 50/10 and does not clear 90/15. So a ninety-minute session on instrumental will loop back on itself, and we would rather write that here than let you discover it at minute seventy. It is a supply problem, and supply is the thing we are adding to.
Two other limits worth knowing before you decide: 90/15 and custom lengths are the paid tier — the free tier is one full 25/5 session a day, which always runs to the end. And only the bundled tracks play with no connection, so the pool numbers above assume you are online.
Related
Study music with no ads — what "no ads" actually means · Why you can't focus while studying at home · Bloc Focus Music