« Open Source - testing a business plan | Main | On generating revenue from Control and Data »

Revenue from Control

In my last article I wrote about basing revenue generation on data and control rather than from selling software.  I recently came across an example of this model – a company called MusicIP .

MusicIP provides an application for cataloguing and identifying music.  My initial view was that is was an iTunes look-alike but on closer inspection there was a twist.  The application (which you can download for free) catalogues your music and then can build play lists of related music – not just by genre or artist but music that is similar in “feel” across all the music you own.  The program also recommends music you can download from the Internet that is also similar in feel. Some of the recommended music is from indie groups who have published their music freely via the Internet as MP3s while other content can be purchased online.

The program does a good job in its recommendations but I wondered how the company could make money and was interested in the possible underlying business models.  Selling music from a third party such as Amazon gets you revenue via affiliate fee but you need large volume to build a business.

Try this rough estimate…  Apple recently passed the milestone of selling one billion tracks of music via iTunes.  Assume the commission rate is 4% (the current Amazon rate) on $1/track sold.  That puts the revenue potential for selling all the iTunes tracks via an affiliate program at $40M.  Double or triple this as you will for other sales outlets beyond iTunes and you have maybe $100M in revenue potential (ok… and growing).  Not bad for a start but not enough to be the only source of revenue.

I dug deeper into MusicIP’s web site and found the diamond in the rough.  They implement a service called MusicDNS that allows you to identify the meta data (artist, title, lyrics, etc.) from the music itself.  The company provides the software to access the MusicDNS service as open source software.

How does it work?  A signal processing algorithm (an FFT) is used to build a unique signature of the music based on the time and frequency distribution of the first 2 minutes (or all if shorter) of the music.  The signature is about 500 bytes long and is sent to the MusicDNS server to look up the music and retrieve the meta data.  Did I mention that you need a key from MusicIP (free for non-commercial users) to access the server?

MusicDNS is like GraceNotes on steroids – if you don’t have the CD identifier or you have a file without embedded meta data, MusicDNS provides a way to access the meta data and monetize the service.

Expanding the use of the service via business relationships could be the basis of a growing business based on transaction fees or annual licensing deals.

It remains to be seen how MusicIP will perform as a business but its an interesting example of a business that appears to pass my Open Source business test (from the last article) and generate revenue from underlying data and control.


The comments to this entry are closed.

My Photo


Intense Brit, lived in Silicon Valley since 1984. Avid pilot, like digital photography, ham radio and a bunch of other stuff. Official Geek.

Proud member of

Venture Capital

a FeedBurner Network

Subscribe to this network

Buy ads in this network

© 2006 - 2009 Stu Phillips

All Rights Reserved.