The content hunter
Martin Stiksel, founder of Last.fm, is back with an even more ambitious project: to organise the entire web according to each user’s behaviour.
Austrian entrepreneur Martin Stiksel scored a huge hit with his music recommendation website Last.fm. His next project, Lumi, goes a step further by delivering all the online content that best matches a user’s interests.
Lumi is based on the same “collaborative filtering” principle as Last.fm. It privately records a user’s Internet browsing history and compares it with that of other users before making recommendations. Users can explore by topic or keyword searches, adding pages to their own personalised, themed collections. Launched in 2013, the service currently has some 25,000 users.
Martin Stiksel spoke to Technologist about the trials and tribulations of a tech start-up – and why he’s doing it all over again.
Technologist How did you create Last.fm?
Martin Stiksel: I was working in London as a journalist for Austrian radio and TV, mostly covering music. In 1999, Felix Miller and I started a platform for unsigned bands to upload their music. We had little experience in programming, and the first website we built was basic. But in a short time we were totally inundated with great music. Then we stumbled across the concept of collaborative filtering: “people who listen to this also listen to that.” People had tried to do music recommendations in lots of different ways – by matching the sound waves of different songs or asking people to list their top 50 bands. Our innovation was to use data that was already available about what music people actually listen to (see The first “scrobbler”). Amazon was doing this for books, but we scaled it up to a much bigger data set – people may read only 10 books a year, but they listen to thousands of songs.
T. Why did you sell the company to CBS Interactive in 2007?
M. S. The whole online music space was becoming quite crowded, and music companies were worried about copyright infringement. Having a large partner in the industry really helped us with that. In fact, online music has become prohibitively expensive to enter as a start-up because record companies want upfront payments for hosting their music.
T. What did you do after you left Last.fm in 2009?
M. S. We took a long holiday. We hadn’t had one in seven years – running a start-up is quite intense. But then we started talking again, and realized that we could build a recommendation system that was not limited to music, which we called Lumi. Other companies collect browsing data on the sly to target advertising, but nobody except Lumi does this for the consumer.
T. So why should I use Lumi?
M. S. It finds things you’re interested in, based on the websites you’re browsing. People sometimes ask, “Why do I need more content? I already have so much!”. But that’s the problem – people face a flood of content on sites like Twitter. Just because you’re friends with someone doesn’t mean you share their music taste or trust their restaurant recommendations.
The more people you follow on Twitter or Facebook, the more unfiltered content you need to prune, and it just becomes completely unwieldy. With Lumi, the more people use it, the better it gets because we’re drawing on the browsing history of like-minded people. If you’re getting into ancient history in Turkey, you can tap into the content viewed by people who’ve already delved into the subject.
T. Has being in the UK helped or hindered your companies?
M. S. It was refreshing to come to the UK, where it’s quite easy to start a company. In Germany and Austria, where Felix and I come from, there’s a lot more paperwork and a lot more investment required.
T. How does it feel to launch another start-up?
M. S. Odd. It’s the last thing we expected to be doing. It’s tempting to leave on a high and not do anything else because people think you’re only as good as your last project. But we also had a success under our belts, so it’s easier to get people to support you.
T. Any advice for budding start-ups?
M. S. You need to find what works by following the data that users generate. Prove or disprove your idea as quickly and cheaply as you can, and then move on.
Also, stick with a good idea. Last.fm took three or four years before we saw any success – and there were some dire times, parents wondering what we were doing with our lives, spending all the money we had saved.
T. What are your hopes for Lumi?
M. S. The Internet may be the best thing that ever happened to humanity, there is so much stuff at our fingertips. But there’s too much great content that goes undiscovered, and we want to change that.
The European Central Bank is offering a payment system to compete with digital services and credit card companies. European start-ups could benefit from it.
Artisense develops a 3D-vision technology for autonomous cars and robots to navigate effectively.