Video: Shazam’s Data Predicts Hit Songs, Identifies Popular Parts Of Tracks

Shazam has been an integral part of the music discovery landscape over the last 15 years, with the smartphone revolution fueling the adoption of the service as the ubiquitous solution to the question “What is the song playing right now?” In the DJ world we’ve watched as the app has also become one of the best tools for trainspotting (picking out the playing track in a DJ’s mix), thanks in no small part to integrating Beatport’s complete music library in 2013 and even adjusting their algorithm so that BPM adjustments made by a DJ don’t impede the tool’s usefulness.

At the Strata+Hadoop  data conference in London this week, Shazam put together an incredible presentation that give us a hint into how they’ve become the best representation of Big Data that we’ve seen in the music industry. The presentation starts off showing how the service’s data can actually do a pretty good job of predicting a hit prior to it getting serious radio play, using the example of a Clean Bandit track.

Identifying the most popular part of a song based on the number of Shazams; Nicki wins.

The rest of the presentation then gets even more interesting for DJs, however. Shazam data can be used to figure out what part of a song is the most interesting to a listener (example below – although who didn’t know that Nicki had the best verse on Monster?) Imagine what this could mean for remixers – instantly being able to identify the most significant part of a song and bringing that into a mix or production.

This part of Clean Bandit’s song is ripe for sampling or mashups.

Shazam can even track the success of a song after it gets remixed – if you live in Europe, odds are pretty good you’ve head the Felix Jaehn remix of Omi’s “Cheerleader” – which in turn helped the original song blow up.

A remix can really make a song catapult into the charts..

One of the best bits was seeing how DJs directly impact the Shazam charts and ultimately influence the popularity of an artist. In the presentation, Shazam’s VP of Product Cait O’Riordan uses Tchami as an example – who has a “spikey” graph that peaks weekly on Fridays and Saturdays when DJs spin his tracks in the club.

Tchami’s success is pretty directly correlated with times in which people are hearing DJs play music.

Watch the entire keynote presentation below:

Comments (20)
Add Comment
  • Shazam Predicts The Hits | MMP BLog

    […] latest innovation to come from Shazam is a kind of digital crystal ball. The company is claiming to be able to predict which songs will […]

  • Jerr1233

    I guess the next question is can we have access to this information or is it stored away and only shared with the labels?

  • Nick Robinson

    Uhhhh by their own chart Bon Iver had the best verse on Monster.

  • couic

    oh that’s great ! a way to know what tracks are commercial crap !!!!

    • CUSP

      Look at it like this: You’ll know which tracks are overplayed and no longer topical. You can use this as a means to be prescient in trimming the fat.

      • couic

        that’s what I meant. except I used harsh language

    • Jerr1233

      This is the thing that I hate the most about being a DJ. Dealing with other DJs who think just because a song has commercial success that its crap. That is elitist BS. You obviously missed the point at the very beginning that in order for them to get data, people must have enough interest in a song in order to pull out their phone and shazam it. Also you missed the point that within the first 24 hours they can start to gauge people’s interest in the song. The 1st 24 hours is before it really starts in rotation on the radio or what not. That means people might actually like a song not because it’s played on repeat and they grow to like it, but maybe, just maybe they actually like the song because they….well….like the song.

      • couic

        oh shit, I forgot to put a smiley to assert that my post was sarcastic.

  • Friday Roundup: The Problem With DJ/Producers

    […] Shazam's Data Can Predict A Hit Tune – Data from music discovery app Shazam can chart how popular parts of a song are, and can also graph the influence that DJs play in an artist's rise to fame. DJ Techtools has the piece Read more […]

  • allday

    The radio here (west coast Canada) is playing the Felix Jaehn remix already!
    Its catchy : )

  • MarcusMCB

    The idea of remixing a song based in part or solely by metrics strikes me as being something of an empty shell. Kinda like ghost-producing or “buying” your social media presence. It all comes off a bit “by the numbers”. Might get you from A to B successfully but where’s the fun in it?

    Funny the article mentions “Cheerleader” as the next possible big hit in the US: it was actually one of the most-added tunes at pop/mainstream radio this past week.

  • Maxey

    Finally remixers can cut out the time consuming task of listening to a track to find the good bit.

    • CUSP

      Giggle… It feels like the world is full of people who want others to do the hard parts for them, and then swoop in at the last second to steal all the glory. Very much in the “Don’t you know who you think I am?!” vein.

  • CUSP

    My Dad deals in data, he’s a business trends forecaster, and he says it’s very important not to read into the data what you want to see, but rather observe the data in many ways to understand not just what the data represents, but why it represents what it does, and always pay attention to contributing factors. This is more-or-less like reading the raw code in the Matrix.

    You can analyze all the data you want, but if the contributing factors to success are no longer present, the kismet isn’t there, which means you can make the best buggy whip in the world, but ever since automobiles were made, the market for buggy whips has dwindled.

    Make sure your market is there. Make sure your product is not tainted. Make sure you can deliver what you promise, and above all, deliver with integrity. Promises are only as good as the belief in them.

    Now, if you’re making music by choosing sections by analytics, you’re not as much an artist anymore, you’re an engineer, or maybe an artist in training, learning from the Masters. You’re not out there in the wild, you’re a trapeze artist operating with a safety net. Sometimes this makes sense, but in order become something great, you must break away from the pack.

    Innovators are revered, Immitatators ride the wave and take credit (including money) for other people’s efforts.

    • BadLarry


      • CUSP

        I guess someone needs a hug so here goes: *Hug*

        • BadLarry

          S0rry br0

  • Mike Steed

    although who didn’t know that Nicki had the best verse on Monster?

    Me…. I don’t even know how it goes!

    • CUSP

      I love how people presume that because something is mainstream that everyone knows about it… it’s pretty arrogant.