Methodology
How the Catalogue Estimator works
The Catalogue Estimator gives a value range for an artist's whole catalogue, across all rights holders, built from public streaming data. It shows a range and a confidence grade, never a single number, because it is an estimate. It is not a valuation opinion and not financial advice.
1. What we measure
For every song we can find, we read all-time Spotify streams, YouTube video views and SoundCloud plays. YouTube Shorts are not counted: a Short is a few seconds of autoplay, not a listen.
2. What we infer
Apple Music, Amazon Music, Deezer, TIDAL and local platforms such as Audiomack and Boomplay do not publish per-song figures to anyone. Their income is inferred from Spotify income, using each market's share of streaming revenue by platform. If Spotify earns a certain amount in a market and holds a known share of that market's streaming money, the rest of the market can be worked out from it.
Where a research firm publishes those shares, we use them (currently US and GB). Everywhere else the share is derived from Roster's own sourced platform weights for that market, adjusted for what each platform pays per play, and it is marked as derived. Derived shares lower the confidence grade.
3. Priced market by market
Plays are split across countries by where the artist's listeners are. The top 15 countries are priced individually and the rest are grouped. Each country gets its own effective rate per stream, because a stream in the United States pays far more than one in Nigeria.
4. From lifetime to this year
Buyers pay for what a catalogue earns now, not what it has earned in total. Lifetime income is converted to a yearly figure on a decay curve: songs earn most soon after release and settle towards a long-term floor. Inside Roster, when the same catalogue has been read twice at least a month apart, the real growth between the two readings is used instead.
5. From income to value
Yearly income is projected forward and discounted to today, with a value for the years beyond the projection. Publishing income is added on top of recording income. A simulation then tests thousands of different decay and discount assumptions, and the range you see covers the likely outcomes, widened further when confidence is lower.
- Streaming decay per year
- 18%
- Long-term floor
- 30%
- Discount rate
- 10%
- Projection
- 10 years + beyond
- Publishing, of recording income
- 25%
- Currency
- Yours, by location
Every figure is worked out in US dollars and shown in your own currency, chosen from the country you are browsing from, at the day's mid-market exchange rate. You can switch to 10 major currencies (USD, EUR, GBP, ZAR, NGN, KES, GHS, CAD, AUD, JPY) with the selector beside the range. Changing currency changes how the number is shown, not the estimate itself.
6. The confidence grade
Every estimate carries a grade from A to E. It says how much of the picture we could see, not how good the artist is. Seven things feed it, each weighted out of 100:
Measured income
The share of estimated income that comes from plays we actually counted (Spotify, YouTube and SoundCloud) rather than from platforms we infer (Apple Music, Amazon Music, Deezer, TIDAL, local platforms) or from the part of the catalogue we could not measure song by song.
25
Hit concentration
A catalogue that leans on one song carries more risk than one spread across many. Worked out from the real songs, so a catalogue is never marked down for a song it does not have.
15
How current income is found
Full marks when this year's income is measured from two readings of the same catalogue at least a month apart. Income worked out from lifetime plays and age earns 40% of these points.
15
Ownership clarity
Full marks when the ownership splits have been confirmed against contracts, half when they have not. The public view never has contracts, so it always earns half.
15
Market reference data
Full marks for markets priced on a published platform share, partial credit where the share is derived by Roster.
10
YouTube fit
How closely the artist's YouTube earnings match what the market shares predict. A big gap means the averages fit this artist less well.
10
Catalogue age
Very old songs are priced on today's audience, which may not be where they were first played. Very new songs are still riding their release spike, so how fast they fade is the hardest part to call.
10
The points add up to a score out of 100, and the score sets the grade. The grade is not just a label: a lower grade trims the central figure and widens the range, so a less certain estimate always gives a more cautious number.
| Grade | Score | Confidence | Central figure trimmed | Range widened |
|---|---|---|---|---|
| A | 80+ | High confidence | No | by 5% each way |
| B | 65 to 79 | Good confidence | No | by 10% each way |
| C | 50 to 64 | Moderate confidence | 5% | by 20% each way |
| D | 35 to 49 | Low confidence | 10% | by 30% each way |
| E | under 35 | Very low confidence | 20% | by 45% each way |
7. What the public view shows
The public Catalogue Estimator shows a range and a grade only, never the breakdown by market, platform or song. It is built song by song from every song we already hold measurements for:
- the artist's 20 most played songs, each with its own plays and release date;
- the full catalogue, song by song, when it has been measured in Roster in the last 30 days;
- the rest of the catalogue, inferred from the artist's all-time totals, exactly as inside Roster.
When the full catalogue has been measured, the public range is the same as the in-app estimate at 100% ownership and default assumptions. When only the top songs have been measured, more of the catalogue is inferred, which the grade reflects. The public grade is never shown above C, because the public view cannot see contracts, deal terms or a second reading of the catalogue. Artists with fewer than 100,000 all-time Spotify streams are not estimated.
The pill beside the range, for example "Confidence C · Full estimate on Roster", shows the grade and leads to the full estimate: inside Roster, rights holders value their own catalogue song by song, market by market and platform by platform, with their real ownership splits and deal terms, and track it over time.
8. Current top 10 earners
The current top 10 earners are ranked, never priced. They are drawn from the artist's 20 most played songs and ordered by what each song earns from streaming right now, not by all-time streams. The list can differ from the most streamed list for two reasons. Songs earn most soon after release and less each year after, so of two songs with the same plays the newer one earns more today. And plays are not worth the same everywhere: a Spotify stream earns more than a YouTube view, and a stream in a higher paying market earns more than one in a lower paying market.
So an artist's biggest song of all time is not always their biggest earner today. A song with a billion streams from three years ago can sit below a newer song that is still climbing.
9. Whose songs count
The estimate values every song in the artist's streaming catalogue, as the platforms list it, across all rights holders. That includes songs where the artist is a featured artist, so it is the value of the music the artist appears on, not what the artist personally owns. Inside Roster, ownership and splits are set for the catalogue being valued.
10. Limits
Inferred platforms are averages, and no artist is average. Radio, live, sync and neighbouring rights income are not in the public estimate. Ownership is assumed to be the whole catalogue across all rights holders, not any one party's share.
A catalogue built mostly on recent hits carries the most uncertainty, because its value depends on how fast those hits fade, and that cannot be measured until time has passed. The catalogue age factor lowers the grade for this, and a second reading a month or more later replaces the assumption with measured growth. Treat the range as a starting point for a conversation, never as a price.