Wow, big news! Instragram has lost 25% of its daily users. A “revolt” declares the New York Post, which broke the story. But numbers seem pretty questionable and perhaps have nothing to do with the terms of service change that upset so many earlier this month.
When I saw the news spreading about Instagram’s drop on Techmeme, apparently a 25% drop in a matter of days, my first thought was — how does anyone know this? Who’s got data on how often I or anyone uses my Instagram account, an application that lives on two different platforms (iOS and Android).
Where The Figures Come From
The data the New York Post used came from AppData, which tracks app activity. Where does AppData get its information? The FAQ page explains:
Data reported from the App Store and Google Play encapsulates rankings for apps and in-game purchases. Because iOS and Android data is provided in terms of rankings rather than actual user counts, we cannot calculate DAU/MAU counts for these apps the way we do for Facebook apps.
DAU stands for Daily Active Users, which is what today’s news is all about. The New York Post reported, according to AppData, that:
Instagram, which peaked at 16.4 million active daily users the week it rolled out its policy change, had fallen to 12.4 million as of yesterday, according to the data.
But the FAQ page says that AppData can’t calculate daily active user counts for Instagram. So again, where’s that data coming from?
Estimating Only From Facebook Logins
AppData has a page that provides more details about Instagram activity here. At the top of that page is this note:
This application is integrated into Facebook from one or more platforms outside the Facebook.com canvas. As such, only users who connect to the app using Facebook are included in the active user counts above & below.
In other words, AppData is looking at data it gets from Facebook about who signs into Instagram using Facebook, as best I can tell. If that’s the case, that’s not necessarily counting much of anything in terms of daily active users. Plenty of people likely log into Instagram without using Facebook.
Why Did The “Revolt” Take So Long?
In addition, look at the traffic chart:
The major drop didn’t happen the day after the new terms were announced on Dec. 17, as you’d expect if there really was a revolt. Of course, the upset about the new terms really came the day after, on December 18. But traffic that day actually actually went up slightly. All those people logging in to pull down their photos? Maybe. But the real drop didn’t happened until Christmas Day and then stayed that way ever since.
What, everyone on Christmas Day suddenly decided, “Instagram, I’m done with that!” If they did, then this other chart from AppData showing that Instagram is the top gaining application would make no sense:
Something else is causing that drop, something that probably has nothing to do with the terms of service. Both TechCrunch and The Next Web note that other apps have seen similar plunges yet had no terms of service change.
Brittany Darwell, from Inside Facebook — which is owned by AppData — also tweeted this to me:
In an email exchange with the Post, an AppData rep noted that the data only represents FB-connected users. Post ignored it.
In other words, it sounds like the Post spotted the drop, AppData may have qualified that with caveats when asked but it got turned into a revolt anyway.
I’m checking with AppData for more clarification, but the company seems pretty good about explaining where it gets data and qualifying what that means on its own site. In this case, those qualifications mean everything.
Postscript: BuzzFeed notes that according to AppData, Flickr has increased slightly, so a sign that Instagram has indeed seen a drop, just one nowhere near as major as 25%:
There’s a much better case for an Instagram exodus, albeit a much smaller one, to be found here: The numbers spiked within a couple days of the backlash, plateaued, and started to settle back down. This is very likely the trail left by disenfranchised Instagrammers.
Indeed, tiny. Flickr went from around 30,000 daily users (versus Instagram’s 16,400,000) to currently 60,000 (versus Instagram’s 12,400,000). So if Instagram lost 4 million users, a whopping 30,000 of those or 0.75%, fled to Flickr.
The reality is that some Instagram users probably did go to Flickr and other services. Many entirely new people probably continued to flock into Instagram. And the Facebook login figures aren’t likely going to reflect any of this well.
Postscript 2: I have this statement now from AppData:
AppData reflects only Facebook-connected users of the Instagram app. Though the terms of service change spurred a lot of negative media attention and complaints from users, the decline in Facebook-connected daily active users began closer to Christmas, not immediately after the proposed policy changes. The drop between Dec. 24 and 25 seems likely to be related to the holiday, during which time people are traveling and otherwise have different routines than usual. A number of other apps saw similar trends, including Skype, Pandora, Pinterest and Yelp.”
From Brittany Darwell, who writes for Inside Facebook, which is associated with AppData, she added this:
We estimate that only 20 to 30 percent of Instagram users connect with Facebook, so there is a large percentage of use that we can’t see trends for. As for the details of how Facebook counts DAU and when, we can’t be positive. There is a three-day plateau for Instagram, but it’s not across the board for all apps, so it’s not a widespread reporting issue. That said, Facebook could stop returning numbers for Instagram if it wanted to. We haven’t discussed this with them yet today. The main point is that AppData is a sample, and to really understand whether proposed policy changes had any effect on the service, we’ll need to observe how that sample changes over time. Data from last year isn’t very representative since so much has changed (not only scale, but percentage of users connecting with FB as well), but I’ll note that Instagram did see a drop on Thanksgiving according to AppData. Facebook-connected DAU was lower on that Thursday than earlier in the week, but the holiday was a record day for number of photos shared.
Darwell also has a short story up at Inside Facebook with more detail, mainly saying that the Post didn’t caveat the AppData information it used very well. It also has a statement from Instagram saying:
We continue to see strong and steady growth in both registered and active users of Instagram.
I’d also asked some additional questions related to how Facebook logins are used to count Instagram users. My questions and responses from AppData:
Question: So if someone doesn’t login using Facebook, they won’t be counted as a daily active user [of Instagram]. Any sense of how many don’t use FB as the login, percentage-wise? Answer: You’d have to ask Instagram for this information. In August, it looked like about 20 to 30 percent of the audience was logging in with Facebook. Question: If someone’s logged in already on their phone, say from a previous day, and they open the app, does that register as a new daily use? Answer: It should, as the app would have to ping Facebook when it wanted to send or receive data from Facebook while open. Question: The key question is what you think happened on Christmas Day to cause the plunge, and why would it stay so long after several days. That suggests that Facebook itself isn’t sending out fresh data. But other thoughts? Answer: None to add; you’re on the right track. Question: I’d love to see the same data for last year, as well…. is that Christmas drop perhaps normal? Certainly others are writing that other apps this year have seen the same. Answer: We can tell you that, yes, Christmas slumps are common for certain categories of regular-use apps because people take a break from their normal app usage habits while on vacation, visiting family, etc.
Comments