Next Big Sound

» see nextbigsound.com as it lived, on the wayback machine

~ the eulogy ~

There is a particular way a good piece of engineering dies inside a larger company, and it has nothing to do with whether the engineering worked. The model gets acquired and keeps running, and then one quarter somebody notices that the reporting layer downstream of it covers most of the same ground for a fraction of the maintenance. The model goes and the dashboard stays. I have watched versions of this happen in a completely different industry and it is the same shape every time, which is why Next Big Sound reads as familiar to me despite my never having had any reason to open it.

It started as a class assignment at Northwestern in 2008. Alex White, David Hoffman, Samir Rayani and Jason Sosnovsky built the first version in White's last semester, raised twenty-five thousand dollars, got into TechStars in Boulder over the summer of 2009 and moved the company from Chicago to Colorado to do it. The site went live in August 2009. Later funding brought in roughly seven and a half million dollars from Foundry Group and IA Ventures among others.

The product was one idea executed well. Artists were leaving numbers all over the internet, on Last.fm and MySpace and Facebook and iLike and Twitter, and every one of those numbers sat in a different place with a different shape and nobody was joining them up over time. Next Big Sound joined them up and plotted them against each other, so you could put one act beside another on the same axes. By the time it had scale it was tracking something like 486,000 artists.

~ the dossier ~

~ what it could actually do ~

The claim that made its name was not a dashboard claim. It was a forecast. Of the 386 artists who appeared on the Billboard 200 during 2014, Next Big Sound had flagged more than eighty percent of them in advance, and a number of those calls came more than a year before the chart entry. On album sales it was quoting forecasts accurate to within twenty percent for around eighty-five percent of the artists it modeled.

That is a serious result, and it held up well enough that Billboard licensed two charts from it and put White on the 30 Under 30 list. The interesting thing about the method is how ordinary the inputs were. There was no insider feed and no label reporting behind any of it, just public counters gathered the same way every day for long enough that the rate of change said more than the number ever did. The signal had been sitting in plain view the whole time. Nobody had bothered to write it down.

~ the acquisition ~

Pandora announced the deal on 19 May 2015 and closed it that July, on undisclosed terms. It made obvious sense at the time. Pandora had an enormous stream of listening data and no external context to put it in, and Next Big Sound had external context and no listening data. Put together they should have been the most complete picture of an artist's trajectory anybody had.

What happened instead was a slow migration into Pandora's Artist Marketing Platform, the free tool set the company gives creators to look at their own numbers and talk to their own listeners. AMP grew, Next Big Sound stayed where it was, and by 2021 the company was describing the shutdown as the final step of a transition that had been running for years. The announcement came on 1 October 2021 and the service went off on 1 November, twelve years after the site first went live, having tracked hundreds of thousands of artists and hundreds of billions of streams.

~ what got buried with it ~

The social data did not move to AMP. That is the sentence that matters and Pandora printed it themselves: the team decided to concentrate on insights that help artists grow on Pandora, which necessarily means the numbers from everywhere that is not Pandora were not worth carrying. So the surviving product tells an artist what happened on one service, and the dead product told them what was happening across all of them.

The Pandora Predictions chart went with it, because it needed both halves to work, and the ability to point at an artist a year out went with the chart. That is the loss. Not a website, which nobody outside the industry used anyway, but a decade-long consistent time series across a dozen platforms, held by one company and switched off by a different one. The old address now redirects to amp.pandora.com, which is the owner keeping the name and pointing it at the replacement, and the replacement can only see its own half.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »