Flurry

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

~ the obit ~

Flurry started in 2005 in San Francisco selling mobile software, much of it aimed at places where the phone was the only computer in the house. That business was fine. The measurement product it built alongside turned out to be the actual company. By 2008 there was an SDK you dropped into an iPhone or Android build and got back session counts and retention curves, at a moment when a developer with no budget had nothing else.

Free was not generosity, it was the design. Flurry never charged the developers, it charged for what a million apps reporting into one place could tell you about the market as a whole. The Pinch Media merger in 2010 pushed its share of mobile analytics to somewhere near eighty percent, and after that the numbers stop reading like a startup and start reading like plumbing: 150 billion app sessions a month, 1.3 billion active devices. Those quarterly State of Mobile reports the tech press quoted for years came out of the same pipe.

Yahoo bought the company on 21 July 2014 for a reported figure above $200 million, in the stretch when Marissa Mayer was acquiring mobile teams faster than Yahoo could absorb them. Flurry became the measurement and targeting layer under Yahoo's mobile advertising, and it outlasted the Verizon purchase along with both renames that followed.

I never installed Flurry and I carried it for most of a decade. That is what this category was. A library shipped inside somebody else's app, counting from a phone in Ankara exactly the way it counted from a phone in Ohio. If you wrote software you knew the name from integration docs. If you only held the phone, there was no screen anywhere that would tell you which of your apps had it in them.

~ the rap sheet ~

~ how it got everywhere ~

The install was four lines and an API key. That is the entire explanation for the market share. Crash reporting and push notifications came later and stayed free, so a solo developer shipping a game in 2011 had no reason to evaluate anything else, and a studio with fifty titles had no reason to run two SDKs. Once the default is set that early in a platform's life it stops being a choice anyone revisits.

~ what yahoo bought ~

The pitch in 2014 was that Yahoo, badly behind on mobile, could buy a view of the entire mobile market and an ad-targeting substrate in one transaction. For a few years it worked well enough that Flurry's data underwrote Yahoo's advertising story to investors.

Then the ground moved. GDPR landed in May 2018 and made passive device-level collection a legal question rather than an engineering one. Apple's App Tracking Transparency in April 2021 put a modal in front of the user and most of them said no. By then Google's Firebase was free and already sitting inside the Android toolchain, which is difficult to compete with from inside a shrinking advertising business.

~ the ninety days ~

The notice went up in December 2023: the SDK, the analytics interface, the programmatic API and the mobile app were all going away on 15 March 2024. Raw data export stayed open until 1 March, and the dashboard held historical data until the final day.

Flurry spent its last weeks negotiating discounts with its competitors on behalf of the developers it was abandoning, and publishing a migration guide pointing at them. Nineteen years, most of the app store, and the last product update was a list of somewhere else to go.

@}-->--

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

« previous  ·  random  ·  next »