Apple Now Lets You Opt-Out of Tracking: Will This Give You More Influence Over Tech Companies?
Apple Now Lets You Opt-Out of Tracking: Will This Give You More Influence Over Tech Companies?
With Apple rolling out some big changes to user tracking in the recent “iOS 14.5” update, you may have seen some fairly staggering early stats: as few as 4% of iOS 14.5 users were letting apps track them. Put another way, app developers just lost access to a large majority of the people they had been tracking across apps — no more personalizing content or ads (or, more importantly, ad revenue).
Apple’s changes have upset the user tracking game by (1) giving users the option to choose whether apps can track their behavior outside the app, (2) forcing apps to prompt users to opt-in, and (3) providing an option to opt-out entirely. The “default effect” — that people typically stick with default settings — is, for now, on the side of tech users who don’t want to be tracked.
The main source cited for the 4% opt-in statistic is an early report from Flurry Analytics, which collects data from a sample of partner apps. Using this sample, Flurry is able to estimate the overall fraction of times any user authorized tracking for any app. Notably, this actually gives more weight to users with more apps installed (the unit of analysis is an “app-user”), so a 4% opt-in rate doesn’t necessarily mean 96% of all users have opted out of tracking entirely. Flurry has been updating their report regularly, and the newer data suggests slightly higher, but still very low opt-in rates of 6% for the U.S. and 16% for the entire world, in line with other reports.
Six percent is still strikingly low, but let’s hold up for a minute. Another analytics firm, AppsFlyer, reported a 40% U.S. opt-in rate and 42% global opt-in rate, using a calculation that focuses only on users who interact with tracking prompts (ignoring those who turned off prompts entirely). In other words, this is a sort of upper bound on opt-in rate, corresponding to a world in which no user opts out of tracking prompts entirely. Of course, even if these higher opt-in estimates are closer to the truth, this still means advertisers lost at least 58% of users in a very short time. Still nothing to scoff at.
These numbers line up with an earlier survey published by Forbes. Several posts have dug into the methodological differences, with a focus on the perspective of firms looking to advertise their products and apps as well as the platforms that publish ads, such as Facebook.
Regardless of which numbers you look at, it seems clear that the level of cross-app tracking on smartphones has already been reduced substantially. And with Android potentially following suit it may take a further hit. But this shift also highlights a new lever for the public to influence tech companies: taking control of their personal data. They could both withhold data from companies with unpopular practices (a “data strike”) or volunteer to give data to companies that have “good” practices (a “data donation”).
Could controllable tracking settings put power in the hands of the public? Could those 6 to 40% of users already opting-in change their minds for a cause? Opting into and out of app tracking is very low cost — as easy as toggling a setting on your phone. This means it’s more likely than ever that a group of people might band together to say “Hey, Tech Company, we will only opt-in if you change a particular practice”. The low cost of changing your tracking settings means following through with this threat would be very feasible, and the public could gain much more meaningful “data leverage” over firms. Critically, Apple has also stated that app developers cannot “gate” features to pressure users to opt-in, which might otherwise compromise potential bargaining power.
Users who opt-out of tracking can be seen as engaging in a “data strike”. As more users opt-out, the impact on machine learning systems accelerates. See more here.
This bargaining power could be quite large. While it’s difficult to generalize about how various firms use the data obtained via tracking, many tracking based systems likely follow a “diminishing returns” curve shared by many statistical and machine learning systems. Losing just 10 or 20 users probably has little impact, but losing more than half of users could put a serious dent in the usefulness of such systems! However, users have another advantage on their side: by opting out of tracking, they cause themselves to receive unpersonalized results, while also harming the ability to train models that personalize ads and content for other users.
While time will tell exactly how the public more broadly reacts to changing options around tracking, it certainly seems that there exists ripe soil for data-focused collective action. Given the low cost compared to other types of protest, would you be willing to tap the “opt-out” button as part of a data strike, and do you think such actions could really have an impact?
