Apple Deprecated Flurry’s Treasure Trove of Data
Back in August of 2011, Apple quietly noted in iOS 5 Beta 6 that it had deprecated the API used to retrieve a device’s unique identifier (called a UDID). There was a firestorm of press when it happened, but Apple didn’t start aggressively enforcing the change until just this week.
For most iOS app developers this isn’t a big deal. Substituting an app-specific identifier will work just fine in most cases. But this seemingly small change is a big deal for ad networks and analytics providers.
Earlier this month Flurry announced “Flurry AppSpot, the first data-powered app advertising platform”. Through its free analytics service, Flurry has collected a staggering amount of data over the past few years, and most of that data is tied to UDIDs. The average iOS user is likely to have several apps installed that use Flurry analytics, which means Flurry is able to use the UDID to correlate the data and build very specific user profiles. With the launch of AppSpot, it’s clear how Flurry intended to use that data — for highly targeted advertising.
Though targeted advertising can quickly turn creepy, I’m not personally opposed to all forms of targeted advertising. In my view, the line between appropriate and suspect ad targeting is in how the data used for targeting was obtained. Unfortunately, much of Flurry’s data was collected by unwitting app developers, like myself, from users who had absolutely no idea any data was collected.
Back in 2009 I installed Pinch Media analytics in Gas Cubby to better understand how people actually use the app. And it’s been a great tool for doing just that, but I was honestly a bit naive in signing up for a free service and not thinking about the long-term business model of Pinch Media. I assumed I’d eventually pay a monthly fee for the service and Pinch would keep iterating on this great tool they had created for developers. Instead they merged with Flurry and kept burning through VC cash while I seemingly got a free ride.
Turns out, I was paying in user data. Instead of offering me the opportunity to pay for the service, Flurry is now going to monetize the data I gave them by offering highly targeted ads. Well, that’s what they wanted to do. By enforcing the deprecation of the UDID API, Apple is rendering useless much of the data Flurry has collected over the past few years. And in my case, I never used Flurry to collect any form of personal and/or sensitive data, so Flurry’s Gas Cubby data wasn’t particularly valuable anyhow.
But Flurry didn’t just roll-over and accept that Apple doesn’t want iOS user data so easily correlated across apps, recent Flurry SDKs seem to capture plenty of other data useful in correlation and Flurry is likely using it to re-correlate the old UDID referenced data. They may even be regenerating the UDID themselves using the the same data used by the deprecated API. Because of this, I’ve decided not to install the new SDK and will completely pull Flurry from my apps in the next update. When I close my account I’ll make sure to request that the data is deleted, but the terms of service I agreed to way back in 2009 likely give them the right to keep the data. Hopefully the data they collected from Gas Cubby ends up in a pool of unusable data and is eventually deleted.
That’s all well and good, but now I have to figure out what to do next — Flurry Analytics was a great tool that helped me make better informed business decisions. TestFlight recently released TestFlight Live which is, in my opinion, much better than Flurry, but TestFlight Live was launched just after TestFlight was acquired by Burstly, an ad network. Same song, second verse. Or is it? This is a great opportunity for Burstly to lead the industry with a clear commitment to developers — and by extension, all mobile device users — about exactly what they will collect, and how it will be used. And if that doesn’t align with Burstly’s long-term business objectives, there’s a big gaping hole in the market for a mobile analytics company that developers can pay for in cold hard cash rather than user data. HockeyApp looks promising.