The Great Messaging-App Migration (That Didn’t Really Happen)
A survey about WhatsApp users’ reactions to the privacy policy update of January 2021
This article summarizes the upcoming CHI 22 research paper “Caught in the Network: The Impact of WhatsApp’s 2021 Privacy Policy Update on Users’ Messaging App Ecosystems” by Carla Griggio, Midas Nouwens and Clemens Klokmose (Aarhus University, Denmark). The paper reports the results from a survey with 1525 WhatsApp users about how they changed their use of WhatsApp and other messaging apps after the announcement of WhatsApp’s new privacy policy in January 2021. In short, 25.97% of participants wanted to switch to other apps (at least partially) due to the update, but only a quarter of those succeeded. Only 0.5% of participants uninstalled WhatsApp. The data shows how installing a new app is easy, but leaving an app is not, mainly due to strong network effects, differences in functionality of different apps, and feeling a loss of control over the distribution of contacts across apps. Get the pre-print here.
In January 2021, WhatsApp updated its privacy policy detailing what types of information it was sharing with Facebook (now Meta), its parent company. Suddenly, millions of WhatsApp users started to install other secure messaging apps such as Signal and Telegram. The influx of new users was such that it even caused Signal to experience short service outages. Back then, it looked as if WhatsApp was losing millions of users to other apps; however, a research study I conducted with Midas Nouwens and Clemens Klokmose at Aarhus University suggests that lots of users tried leaving WhatsApp, but actually, most failed. In this article, I share some of the main results from this study.
What did WhatsApp’s new privacy policy change? It mainly provided more details about how WhatsApp was already processing user data, including data they share with other Meta companies (e.g., Facebook, Instagram, Messenger). This was clarified by Will Cathcart, Head of WhatsApp at Meta:
For example, the previous policy stated that they collected location data only when using location-based services, but the new one adds that they also use IP addresses and phone number area codes to estimate users’ location, even when location-based services are not used. The new policy also clarifies what information is shared with other Facebook Companies, which may include the user’s phone number, transaction data (e.g., from Facebook Pay), interaction data (e.g., how often they use features or communicate with businesses), device information, and IP address. While the European Region version states that the shared data can only be used on WhatsApp’s behalf (e.g., for infrastructural services), the non-EU version says that it can also be used for the benefit of other Facebook Company Products (e.g., showing relevant ads). The new privacy policy also states that WhatsApp business accounts could store their chats on Meta’s servers to access marketing services, including advertising on Facebook.
WhatsApp announced this new privacy policy by showing a pop-up message to all users, asking them to accept the new terms and privacy policy by February 8 in order to continue using WhatsApp. This caused an immense commotion worldwide, and in less than a week, Signal saw 7.5 million downloads, and Telegram registered 25 million new users.
Meanwhile, WhatsApp tried to clarify that messages were still private and secure. As millions of users kept installing other apps, WhatsApp first moved the deadline for accepting the new privacy policy to May 15 and eventually decided that users could continue using WhatsApp even if they didn’t accept the update.
From what we could see in the news and social media, this was an unprecedented “app” migration, and we thought it was a unique opportunity to study how users choose the messaging apps that they want to use in response to an update that they didn’t want.
We launched a survey soon after what would have been the first deadline for accepting the new privacy policy. We asked participants what other messaging apps they used besides WhatsApp, how often they used them, and what types of relationships they had with the contacts in them. We also asked them whether they changed their use of WhatsApp and their use of other apps since the update announcement, for example, whether they uninstalled WhatsApp or used it less often, or whether they installed new apps or used other apps they already had more often.
In May, soon after the final deadline for accepting the update, we distributed a second survey to the same participants, resulting in a total of 1525 WhatsApp users from Mexico, Spain, South Africa, and the United Kingdom. We asked the same questions again to see if there were changes in the apps they used between February and May, and we asked additional questions about their response to the update. In particular, we asked if the update made them want to switch from WhatsApp to other apps and if they had managed to switch as much as they wanted. We also asked about what apps they tried as an alternative to WhatsApp, why they chose those apps, and what challenges they faced while trying to switch.
In this article, we focus on the data we collected in the second survey (May), but in our upcoming CHI 22 paper, we include insights about the comparison between February and May. Our results offer nuanced evidence of how challenging it is to switch messaging apps, showing that installing a new app is easy, but leaving an app is not:
Participants increased their use of other apps but also kept using WhatsApp
First of all, we found that because of the update, 25.97% of participants wanted to switch to other apps (at least partially), but only a quarter of those, 6% of all participants, felt that they managed to switch as much as they wanted. 24.92% of participants explicitly tried other apps as an alternative to WhatsApp, but only 0.5% uninstalled WhatsApp. Let me put these numbers into perspective: if we translated this result to the entire population of WhatsApp users (approx. 2 billion), we’d be talking about 500 million users trying to flee from WhatsApp, and only 10k uninstalling it.
27% had increased their use of other apps, either by installing new ones or using the ones they already had more often. However, only 16% reported using WhatsApp less, and merely 8 participants uninstalled it.
All participants used WhatsApp, but it wasn’t the only app they used. They used a median of 5 apps in total and 4 apps regularly, including one “primary” app that they used very frequently. For 78% of participants, this primary app was WhatsApp.
Now, despite using a median of 5 apps, participants reported an average of only 6 contacts with whom they chatted across multiple apps — mainly romantic partners, friends, and relatives. Overall, they communicated with a median of 2 types of relationships per app, but, the median for WhatsApp was 6. Previous research talks about how people distribute different contacts across different apps as a way of drawing boundaries between them. However, here, it looks like WhatsApp is the place to chat with anybody, making it particularly hard to leave.
Another finding that points to how hard it was to leave WhatsApp is that the participants that felt satisfied with the extent to which they managed to switch did not just try to move to one other app — they tried a median of 3 apps, either by installing new ones or using the ones they already had more often.
Deciding where to switch to was a complex task
A third of the participants trying to switch identified that it was challenging to agree with different contacts on an alternative app. Some explained this challenge in their own words:
[The] extra task of having to remember which contacts are in which app.
Trying to move all your contacts to find out that most of them are not moving to the same app you’re using.
The most common apps chosen as alternatives were Telegram and Signal, mostly because of their privacy and security features. However, almost a third of the participants that tried to switch chose to move to either Facebook Messenger or Instagram, which we find sadly ironic. Both apps are also owned by Meta and don’t offer end-to-end encryption by default. Their privacy policies explicitly say that they can use the content of conversations for improving their products, or users’ location for targeted advertising. If participants’ intentions behind fleeing WhatsApp had anything to do with the user data they shared with Meta, this should be taken as a red flag about how hard it is to make informed choices in the messaging app market. Participants reported diverse challenges related to choosing alternative apps:
Reading through the terms and conditions of the new apps.
Researching the different apps and choosing one based on security, functionality, and design.
There are no other apps that do the same thing.
Breaking Network Effects: Messaging Interoperability and Beyond
We believe that the main source of these barriers comes from the forced symmetry between messaging apps, that is, that the users of an app can only message other users of the same app. We believe that a promising solution is to enable some degree of interoperability so that users can message anybody they want regardless of the apps they use. To make this happen, we think that a multidisciplinary approach is necessary, where HCI researchers, designers, and privacy & security experts work together with policymakers.
Many countries and regions are working on regulating competition in digital markets, requiring platforms to “open up” and offer interoperability with competitors. For example, in 2018, the European Union reformed the European Electronic Communications Code, a directive that, among other things, encourages “interoperability of electronic communications services” to lower network effects and promote fair competition. While this directive would apply to messaging apps, it isn’t specific about what types of functionality should be interoperable, and it doesn’t require interoperability as a strong obligation. WhatsApp published an extension to their terms of service in response to the EECC but included no mention of messaging interoperability. On the other hand, in December 2021, the European Parliament approved an amendment to the Digital Markets Act, a draft law for regulating competition in digital markets, which included explicit requirements for messaging apps (i.e., number-independent apps) to offer interoperability:
In the particular case of number-dependant intercommunication services, interconnection requirements should mean giving the possibility for third-party providers to request access and interconnection for features such as text, video, voice and picture, while it should provide access and interconnection on basic features such as posts, likes and comments for social networking services. (Digital Markets Act, version of 18/11/2021)
Our study contributes new evidence that network effects hurt freedom of choice in the messaging app market, and we hope it will help push such regulation forward where messaging interoperability is a requirement.
However, what specific functionality needs to be interoperable is still an open question that calls for research on how interoperability affects interpersonal communication. Cross-app messaging would require a standardized protocol, and the question is: what type of communication functionality should that protocol include? We found that beyond network effects, many of the participants that tried to switch were frustrated about losing the functionality and means of expression they had in WhatsApp. The functionality that users adopt for their online expression can be quite personal and diverse, making it challenging to define a “minimum common denominator” of interoperable functionality.
Participants also struggled with disagreements with other contacts switching about what new app they should connect through and with finding contacts in the new apps. How would users control which contacts can reach them on each app, if suddenly it was possible to contact a person on any of their apps? Moreover, beyond a standardized protocol, how would asymmetries in the functionality of apps affect communication? For example, what would happen if an app is designed to delete messages after 24 hours, but the other one saves all conversation histories? What if one app supports animated stickers and reactions that other apps can’t show?
If you’re curious about interoperable messaging, I recommend trying out Element.io or any app built for the Matrix.org protocol, as well as Beeper.com. And if you have experience with interoperable messaging or are interested in the topic, feel free to reach out!
