Facebook's iOS Glitch Caused ComScore to Overestimate Time Spent
Last fall people discovered that Facebook's iPhone and iPad apps were draining their devices' batteries because the apps were making calls to Facebook's servers even when the apps were running in the background, making the devices act like someone was using the app even if it didn't appear on their screen.
Those background calls didn't only drain devices' batteries, as it turns out, but also made it seem like people were spending more time using Facebook's iOS apps than they actually were.
Facebook fixed the issue in October 2015 by reducing the number of background calls its iOS apps make. Months later the Video Advertising Bureau asked digital measurement firm ComScore -- which uses calls made by Facebook's apps as a proxy for time spent -- to see if those background calls muddied measurements of how much time people were spending in Facebook's iPhone and iPad apps. Based on ComScore's measurements comparing time spent in the month before the fix and the month after it, the glitch in Facebook's iOS apps appears to have inflated the total amount of time people in the U.S. spent using Facebook's mobile apps by double-digit percentages.
The Video Advertising Bureau isn't exactly an impartial party. The trade organization formerly called the Cabletelevision Advertising Bureau is backed by TV networks and pay-TV providers. These companies are, at best, frenemies with Facebook. At the same time as Facebook serves as a way to promote TV shows and measure how audiences feel about a show, the social network is also competing against the networks for the advertisers' budgets and audiences' attentions, particularly as Facebook positions itself as a dominant digital video service.
The group takes its cues from the interests of marketers and agencies, according to VAB President-CEO Sean Cunningham. "They inform our agenda of what the topics are and what they want numbers on and what they want us to keep an eye on," he said.
Over the past five quarters, he said, the organization has been focusing on time spent because it's a comparable metric that could be applied across different channels, such as TV and digital, as opposed to metrics like view counts that can vary even between individual platforms like Facebook and YouTube.
Now for the numbers.
According to figures that the Video Advertising Bureau received from ComScore, and that ComScore confirmed to Ad Age, people in the U.S. spent 23% less time using Facebook's mobile apps in November 2015 compared to the amount of time they spent in September 2015 before the glitch was fixed. And the average amount of time spent by each person who used Facebook's mobile apps was 24% lower in November than in September.
When looking at time spent on Facebook across its desktop and mobile sites and mobile apps, the total amount of time spent on Facebook was 19% lower in November compared to September, according to ComScore. And the average amount of time each person spent on Facebook across those channels was 20% lower.
When looking at Facebook's iPhone app specifically, total time spent was 40% lower in November compared to September, and the average amount of time spent per person was 41% lower. For Facebook's iPad app, total time spent was 39% lower, as was the average amount of time spent per person.
For comparison, total time spent in Facebook's Android app increased by 2% and average time spent per user was flat when comparing September and November; ComScore's Android figures are considered more reliable than its iOS figures because the firm is only able to take into account activity when the app is running in the foreground.
Facebook initially declined to comment. In a statement after this article published, a spokeswoman said, "Just last week we mentioned on our earnings call that we grew both the number of people using Facebook on mobile, and time spent."
The double-digit discrepancies are eye-opening, but given Facebook's size and the way advertisers spend their money, at this point in time they are more of a cautionary tale than a controversy.
"Yes, it's not good that there was a glitch," said Jeanne Bright, VP and director of paid social and strategy at DigitasLBi. "And yes, it's not good that there was such a big discrepancy in that time period. But Facebook still has such a huge amount of reach, and I mean all of their properties [including Instagram and WhatsApp]." Last week Facebook reported that mobile accounts for 90% of the more than 1 billion people that check out the social network each day.
Time spent is generally used by advertisers directionally if at all, and even if time spent on Facebook were overestimated, people still spend lots of time on Facebook, said one agency exec who asked to remain anonymous. Even though individual users in the U.S. spent on average 41% less time using Facebook's iPhone app in November 2015 than they did in September 2015, the average person still spent 636.2 minutes using Facebook's iPhone app in November.
"It doesn't really make a huge impact, the time spent," Ms. Bright said. "From a general media planning perspective, I think people more so look at uniques, which obviously Facebook just crushes it. From a buyer's perspective, it certainly helps that people are spending so much time in Facebook because that means you have even more opportunities to reach people. But it's more like a fun statistic we like to put into decks to impress our clients, but not something we use as a basis for how much money Facebook gets on a direct basis."
Kieley Taylor, who heads GroupM's paid social practice, concurred. She said that time spent is not one of primary considerations when WPP's media-buying arm is deciding how clients should spend their money. "Our approach takes into account the cost effectiveness and how many people we can talk to across platforms as an apples-to-apples metric," Ms. Taylor said. "In that sense, dollars weren't inflated going into the platforms. The rationale, perhaps, is just not quite as strong for why we would have the dollars on Facebook as mobile-first, maybe."
As time spent becomes a more important metric to ad sellers like Facebook and ad buyers like DigitasLBi and GroupM, however, it becomes more important for advertisers and agencies to be aware of how reliable those numbers are. "We're encouraged that it was caught before it had become the new currency," Ms. Taylor said.
Facebook has stopped using these calls to calculate its internal time spent-figures, which the company likes to pass on to advertisers. "When Facebook is selling their platform unsubstantiated directly to our clients, which happens a lot, those are the types of metrics that they really like to showcase," Ms. Taylor said. But ComScore has not adjusted its own methodology. That means the measurement firm's figures for time spent in Facebook's iPhone and iPad apps is still susceptible to any background calls that Facebook's app makes, even if Facebook has reduced the number of those calls.
Compared to tech companies' internal numbers that some advertisers and agencies are able to access, most third-party measurement firms struggle to provide accurate data about social networks as well as search engines, according to one agency who asked to remain anonymous. But that doesn't mean advertisers and agencies would prefer to rely on Facebook's own numbers. "We still think that third parties' broad strokes are better than folks grading their own homework," Ms. Taylor said.