How to Read Your Listening Statistics
Listening statistics are only interesting once you know what is being counted. This walks through each figure Melodius shows you, what it is derived from, and what it does and does not tell you.
Every figure on a stats page comes from the same raw material: a stream of play events, each recording which track was played, when, for how long, and whether it finished. Everything else is an aggregation of that. Knowing which aggregation you are looking at explains most of the surprises.
Pick the period first
Melodius computes stats over a period you choose — the last 7 days, 30 days, 90 days, a year, or everything. This matters more than it looks, because most figures are also compared against the equivalent previous period, and a short window is volatile.
The overview numbers
- Listening time
- Total milliseconds of audio actually played, not track duration summed. If you skip halfway through a six-minute song, three minutes are counted. This is why listening time and play count can move in opposite directions.
- Tracks played
- The number of play events, not the number of distinct songs. Playing one song forty times gives you forty plays. Compare this against distinct tracks to see how repetitive a period was.
- Distinct artists and distinct tracks
- How many unique artists and unique songs appeared at all. The ratio between distinct tracks and total plays is the single most informative thing on the page — it is the difference between a wide month and a deep one.
Each of these shows a percentage change against the previous period of the same length, so a 30-day view compares against the 30 days before it.
Top tracks and top artists are ranked differently
This trips people up. Top tracks are ranked by play count; top artists are ranked by time listened. They are not the same measure and they can disagree.
An artist whose songs run seven minutes each can be your top artist on time while never appearing in your top tracks, because no individual song accumulated enough plays. An artist with one three-minute single you played constantly can dominate your top tracks and rank low on time. Neither ranking is wrong; they answer different questions.
The heatmap and the hour distribution
The heatmap shows daily listening across roughly the last year, bucketed into intensity levels. It is not there to be precise — it is there to show shape. Gaps, streaks and seasonal patterns are visible in it that no single number captures.
Alongside it, the hour distribution shows when in the day you listen, plus your most active hour and most active weekday. This is the part of the stats page that most reliably tells people something they did not know about themselves, usually that their listening is far more concentrated in one part of the day than they assumed.
Sessions
A session is a run of listening with no long break in it. Melodius treats a gap of about half an hour as the end of one session and the start of the next, and reports how many sessions you had, the average length, and the longest.
Session data separates two very different listening lives that produce identical totals: ten hours as one long stretch a week, versus ten hours as thirty short bursts. The totals match, the experience does not, and only session length shows it.
Listening style: the four ratios
Discovery rate
- What it measures
- The share of your listening that went to things you had not played before
- How to read it
- Low is not bad — it means you are deep in a phase. Consistently near zero across a year means the rotation has closed.
Repeat rate
- What it measures
- How much of your listening was replays rather than distinct tracks
- How to read it
- High repeat with high discovery is a healthy pattern: finding things and then actually living with them.
New artists
- What it measures
- Artists with no plays in the previous period who appeared in this one
- How to read it
- The most honest single measure of whether you are actually widening out.
Genres explored
- What it measures
- How many distinct genres appeared in the period
- How to read it
- Useful as a trend rather than an absolute — genre tags are coarse and inconsistent between catalogues.
If discovery rate is the number you want to move, the techniques in how to discover new music target it directly.
On Repeat, New Discoveries and Taste Evolution
- On Repeat surfaces the tracks with the highest replay counts in the period — the songs you are currently wearing out, which is often not what you would have guessed.
- New Discoveries lists artists with no plays in the previous period. It is the concrete version of the discovery rate percentage.
- Taste Evolution compares artist play counts against the previous period and splits them into rising and declining. This is the closest thing to a view of where your taste is heading rather than where it has been.
Music DNA and its evidence
Melodius classifies your listening into an archetype — Explorer, Loyalist, Night Owl and so on — with a secondary archetype alongside it. Usefully, it also shows the evidence: your completion rate and your skip rate, which are two of the signals the classification is built on.
The classifier needs a reasonable amount of listening before it says anything meaningful, so a new account will not have one straight away. The full list of archetypes and what triggers each one is worth reading before drawing conclusions from yours.
What stats cannot tell you
- Whether you enjoyed it. A high play count on a track you left running while working is indistinguishable from one you loved.
- Anything about listening that happened elsewhere. Stats only cover plays on this account.
- Genre with real precision. Genre tags come from catalogue metadata and are inconsistent — treat genre counts as directional.
- Causation. Your listening went up in March; the stats do not know that March was when you started commuting again.
Frequently asked questions
Why are my top artists different from my top songs?
They are ranked by different measures. Top tracks are ordered by play count, top artists by total time listened. An artist with long songs can lead on time without any single track reaching your top songs, and vice versa.
What counts as listening time?
Audio actually played, not the full duration of the tracks you started. Skipping halfway through a song counts only the part you heard, which is why listening time and play count sometimes move in opposite directions.
What is a listening session?
A continuous run of listening with no long gap. Melodius treats a break of roughly half an hour as ending a session, then reports how many sessions you had, your average length and your longest.
What is a good discovery rate?
There is no target. A low discovery rate simply means you are deep in a phase rather than sampling. It is worth attention only when it stays near zero across a long period, which suggests the rotation has closed rather than settled.