The Evolution of Super Apps

Team Members

Fernando van der Vlist, Anne Helmond, Michael Dieter, Francesca Mauri, Karla Zavala Barreda, Marco Giacomazzi, Martin Trans, Natalie Kerby, Bakar Abdul-Rashid Jeduah, Samir van Oeijen Rodríguez, Sara Ferrini, Yan Cong, Silván Waidmann Irastorza, Rajender Singh

Project facilitators: Fernando van der Vlist, Anne Helmond, Michael Dieter

Image: https://twitter.com/appstudies/status/1545450527118475265. Posters are attached to this page (below).

Introduction

In this project, we empirically explore what constitutes a "super app" and how such super apps evolve over time and geographically. The project builds on previous app studies projects aiming to develop approaches for examining mobile apps and app evolution (see Dieter et al., 2019; Helmond and van der Vlist, 2021). It aims to contribute to existing research on super apps (Chen et al., 2018; Goggin, 2021; Steinberg, 2020) by dissecting what super apps are and how they are shaped by their local cultural, political, economic, and regulatory contexts, resulting in varied types of super apps evolutions across different geographies.

Data Sets

This study was conducted with four data sets. The first was created from a corpus of super apps drawn from an exploration of expert lists available online, as well as apps that call themselves super apps. This dataset includes 44 super apps, their original features, their current features, release dates, regions where they are active, as well as their google play category. The second dataset consists of the super apps which have a family of apps; it shows the apps which form part of the family and the ones that have been deleted. The other two datasets consist of the list of top 100 third-party apps that were built for TikTok (3rd dataset) and WeChat (4th dataset). We queried the Google Play store to retrieve these lists on a weekly basis between 2020 to 2022.

Research Questions

Research question(s): What defines a super app and how do they evolve over time and geographically?

Sub-questions:

• Which different kinds of super app types exist?
• How many and what kind of features do super apps offer?
• How do their development methods compare in terms of their own mini-apps and supporting app clusters?
• How are the different types of super apps distributed around the world and can we see regional tendencies and cultural entanglement?
• How did super apps evolve over time?
• How does the market influence the direction of super apps?

Methodology

Methodology for data collection

The data for this project was collected by identifying 44 super apps through internet search. We used Google to search for super apps in different regions of the world. Once we had the list of apps, we followed the steps below to collect more details about each of them:

We first compiled a list of the super apps’ app IDs from data.ai and the apps’ Google Play Store URLs. We then retrieved data (like description, release date, number of downloads etc.) about them from the Google Play store using the ASI App Store Scrapers, setting language to English and country to US. We also manually identified the Google Play Store categories for each of these apps and searched to see if they have any micro-programs that allow third party developers to engage with these apps. We read through the play store description of these apps and also checked their websites to categorize their services under different themes. We engaged in an emergent coding process to identify shared features across the different apps, and identified 18 shared features between the super apps collected. We then proceeded to unify said features under eight overarching common functionality areas: Payment & Finance; Shopping; Gig Economy; Social; Booking; Health & Fitness; Entertainment; Utility (See table 1).

This project has affirmed that much effort must go into establishing common standards and categorisation schemes if typologies are to be derived from very fine analysis of sets of features, as an example. Since for several reasons (including language barriers and geoblocks on top of time restrictions) it was unfeasible to install every single app we, as a group of diverse researchers, were relying on a mixed set of sources such as individual prior experience, journalistic and commercial publications, app store descriptions and app-platforms such as data.ai. With categorisation being a shared process across all ten participants, many conversations spurred a continual change of the granularity and specificity of the ad-hoc enabled macro and micro categories.

Overarching categories	Feature	Definition
Payment & Finance	Banking/Financial	These apps offer money storage, loans, they provide credit and or debit cards along with other financial services (i.e. insurance)
	Payment	These apps allow users to make payments but don’t have an internal financial storage system. They allow users to pay bills, exchange money and deal with other forms of transfer.
Shopping	Rewards	These apps offer rewards on shopping or expenses in the forms of cashbacks or deals.
	Marketplace / E-Commerce	These apps have an internal shopping service and act as platforms for both users and businesses to display their goods.
Gig Economy	Food Delivery	These apps allow users to order meals from restaurants and other businesses. It distinguishes between other forms of food delivery because the delivered goods are prepared to be consumed.
	Package Delivery	These apps allow users to receive any kind of package, from regular mail to groceries of any kind.
	Ride Hailing	These apps offer ride hailing or ride sharing services.
	Microtasking	These apps provide HAAS (humans as a service). One can find cleaning or hairdressing services, as well as people who do specific micro-tasks (i.e. retrieving a pair of forgotten keys).
Social	Chat/Messaging/Calls	These apps offer a different range of communication services, from chatrooms to calls and video-calls among single users or groups of users.
	(Social) Content	These apps act as a platform for users to produce, share and display their content, which then can be accessed by circles of friends or publicly.
Booking	Events/Restaurant/Movies	These apps allow users to book tickets to a variety of venues, from restaurants to clubs to cinemas.
	Travel	These apps allow users to book tickets or other traveling services.
Health & Fitness	Health & Fitness	These apps allow to monitor users’ health and to organize the
Entertainment	Series/TV/MusicStreaming	Hosting and accessing the content on the app. These apps allow the user to either watch television series, listen to music or watch TV, or do some or all of these, through the app.
	News	These apps offer news services, sometimes producing their own news through internal editorial boards.
	Games	These apps have built-in games.
Utility	Work/Office	These apps can be used as tools to facilitate teamwork or to produce different kinds of documents.
	Cloud/Data Storage	These apps offer cloud storage for users to store and share their data

Table 1: Eight overarching categories, eighteen features and their respective definitions derived from the emergent coding process.

Methodology of Super App Typologies

We developed a typology of super apps and built a methodology to categorize and analyze how our collected super apps fit into this typology. We did this by identifying the relationship between app features, whether it had supporting apps, features arising from first or third-party developers, or if it belonged to a family of apps from the same developer.

As Goggin observes (2021), “the super app concept does not really pass muster analytically... [yet] it is useful to explore and unpack it, as it offers insights into the economic and industrial trajectories of apps." Given the competing conceptions and definitions of a super app that circulate within industry literature, we devised a typology to clarify common development strategies identified in our dataset. These include: the app family as a set of apps developed by a single company over time, often with an overarching brand identity (i.e. Uber, Uber Eats: Food Delivery, Uber Freight, Uber Bus, etc., or Facebook’s ‘family of apps’ see Nieborg & Helmond, 2019; van der Vlist et al., 2022); the app ecosystem as a co-evolving environment of auxiliary or ‘support’ apps developed by third-parties for a single dominant app (i.e. ‘regramming the Instagram platform’, see Gerlitz, et al. 2019); the swiss army knife as the integration of multiple first-party services and features into a single app (i.e. MyJio or Line, see (Steinberg, 2020); and, finally, the mega-platform where third-party ‘mini-programs’ or complementors are developed and hosted ‘inside’ a single app (i.e. WeChat, see Chen et al., 2018).

The different typologies are not mutually exclusive. Therefore, we built a methodology to code every app according to one or more typology (see table 2).

Table 2: Criteria for inclusion of super app in super app - type

Methodology for App Family History

To trace and visualize the history of App Families, we chose to analyze the development in organizational structure of 3 super apps with similar origin points, i.e. Gig Economy. Our planned approach was to collect a significant subset of APK releases, to then analyze changes in feature, structure, trackers etc. However, methodologically, collection of accurate timelines of version updates for analysis of the software proved to be unfeasible. Additionally, feature updates were not always adequately reported by app developers. For this reason, we decided to shift our focus to the development of App Families over time.

Having decided on a new direction for historical analysis, we retrieved the ‘family of apps’ from all our 44 super apps using the app analytics service data.ai. For the complete set of apps, we collected their launch dates and store removal dates from data.ai. We collected apps published by the same app developer on the Google Play Store, excluding other app stores to attain uniformity in collection, and extracted publishing dates of the different members of the families of apps. We also included deletion dates of family members. In this process, data of removed apps of LINE was accidentally omitted, which was acknowledged in the visual representation of the results.

Findings

App Categorization

Examining a super app’s category in Google Play offers an initial understanding of which feature of the super app is seen as the primary feature by the app. When app developers publish apps to Google Play, they have to choose an app category which “help[s] users to search for and discover the most relevant apps in the Play Store” (Google). Although super apps provide a range of services, the Google Play categories reveal what super apps present as their core business, and may also indicate the initial purpose of the app. The app ID (e.g. “com.grabtaxi.passenger”) can equally convey information of an app’s intended purpose as it cannot be changed after publishing, potentially giving off information of the development of the app and journey from app families to consolidated super-apps.

In our analysis of the collected data, three main clusters emerge: Finance, Maps & Navigation, and Communications. The three main clusters coincide with Ajene’s (2021) super app archetypes. Even though this categorization was first thought to describe the super app ecosystem in Africa, it still holds to our scoping of super apps globally. For instance, the Maps & Navigation category correlates with Ajene’s mobility-driven apps, the Finance category corresponds to Financial Services-Driven, and Communications to Telco-driven.

App Features

These overarching categories and features allowed us to explore which shared features super apps have. Could we use the number of different features across overarching categories combined in a single app as an indicator of ‘super-appiness’? Could we trace their historical emergence and the evolution of apps from simple to ‘super’?

The division of features also allows conducting other types of historical enquiry, including comparisons of apps’ initial features and the ones acquired later over time, though this will require further work in establishing a standardized information gathering scheme to accomodate for the various sources of this information, such as news outlets, tech blogs, changelogs and the APK files themselves.

In our visualization, three main clusters of features emerge: Payment & Finance, Gig Economy, and Shopping. This reveals that there are three features that are the most prevalent among ‘super’ apps, with Payment & Finance features typically being the initial feature that supports the expansion into additional feature types. It shows that the rise of super apps go hand in hand with the rise of digital payment and fintech. This also explains why shopping is one of the main features super apps offer, even though it was not a major category in our above finding.

Theory of Super App Typologies

See --> Methodology of Super App Typologies

Atlas of Super Apps

In our atlas, we plotted 14 apps for a proportional geographic representation of the distribution of super apps across the global landscape, taking into account the apps’ superness, original features, and number of downloads.

Super apps with the characteristics of the swiss army knife model mostly originate in African and Asian markets, and are active in these respective regions. While WeChat also has swiss army knife characteristics, it also functions as a mega-platform, since it has mini-programs, and has an app ecosystem around it, making it uniquely structured in this region.

Apps with headquarters in the United States, such as Google and Uber, tend to follow the “app family” model and be active worldwide. Facebook slightly differs from the other two, considering it not only has an app family, but also functions as a mega-platform, swiss army knife, and has an ecosystem of third-party apps. In this sense, it is somewhat closer to WeChat —even in the fact that they both developed out of the same overarching category of social apps—although they originated from different cultural and geographical contexts.

The geographical distinction between super app models may be due to geopolitical and legal reasons, such as the GDPR and heightened focus on anti-trust legislation in the EU and US. We consider this a point for further research.

The number of downloads an app has on the US Google Play Store determines the size of the squares in the visualization. Generally, these numbers were consistent with local Google Play Store downloads. It should be noted, however, that the amount of app downloads are heavily influenced by Android being open for factory preinstalled software (bloatware), as the Google app comes preinstalled on virtually every Android phone off of the assembly line. Facebook, along with some other apps, is commonly seen preinstalled on Android phones in various regions of the world (Gamba et al.), understood along the lines of a sub-practice of Zero Rating (Hoernig & Monteiro). This is, naturally, somewhat eschewing the visualization.

Due to missing data, the size of WeChat is not representative of its actual global usage. The number shows the downloads of WeChat’s international app hosted by the Google Play Store, rather than from relevant Chinese app stores.

History of App Families

We looked at the development of App Families over time in search of common patterns. This was done by collecting apps published by the same app developer on the Google Play Store, excluding other app stores to attain uniformity in collection, and extracting publishing and deletion dates of the different members of the families of apps. The resulting data collection allowed us insight in the evolution and expansion strategies of family apps. Visually, bars represent the time a family member was active.

Supporting Family: A group of apps was identified by their growing supporting app family group throughout time. Facebook (Meta), Didi and Uber have family groups that support different features of their business models and are aimed at different stakeholders.

Uber on the other hand, shows both App Family structures. It’s family includes apps that provide distinct services to consumers (e.g. Uber, Uber Eats, Jump), and those that support existing services (e.g. Uber Fleet, Uber Driver, Uber Eats Manager).

It is similar to Didi’s App Family, except that Didi’s expanding app family also reflects the app’s global expansion. Between 2018 and 2022, Didi released separate ride hailing apps in Japan, Russia and the UK.

With Facebook we can see that a lot of their supporting apps are aimed towards important customer segments and provide additional functions for them. It would be interesting to analyze how much Facebook’s first party apps are used now. If for some of the apps there is a low user number it poses the question why Facebook is keeping them up. Facebook seems to put thought in their release date and release apps in incremental time intervals.

Amazon, Google, VodaPay, and PayTM all have deleted multiple apps over the years. According to the typologies, each of these apps operates as an app family, while Amazon also has an app ecosystem and VodaPay has characteristics of a mega platform and swiss army knife. While VodaPay ’s trajectory suggests that these apps consolidated into one main app, Amazon, Google, and PayTM likely have many deletions considering they rely on first-party development and thus, experiment and iterate on their app designs. This could be regionally and therefore legally tied, considering Amazon and Google are US-based companies, VodaPay in South Africa, and PayTM in India.

The Evolution of App Ecosystems

App ecosystems consist of third party apps that get built around a particular app with the objective to enhance the functionalities or outputs of a particular app. For example, there are hundreds of apps built by third party developers for snapchat. We followed the following steps to check if a particular app has an ecosystem of apps:

• We searched for the keywords in a target app's name in Google Play Store.

• We examined around top 20 results to see if we have any third-party apps with names or descriptions that contained the keywords in our target app.

App ecosystems consist of third-party apps that are built ‘for’ other apps. We queried the Google Play store for [!TikTok] and [!WeChat] on a weekly basis between 2020–2022 and retrieved the top 100 apps for these queries. We observe how the ecosystem of apps around TikTok and WeChat evolved over time. While for WeChat 15.76% of the apps were removed from the store, for TikTok this is 56.94%! TikTok ’s ecosystem thus represents a ‘spammy neighborhood’ corresponding with its quick rise in popularity in the same period. Finally, the top of both app ecosystems is fairly stable, while the long tail is very volatile. This is a unique image of an evolving dynamic ecosystem, warranting further research.

Discussion

Informed by both the app categorization and the shared app features from the collected data set of super apps, we identified super app typologies: a) family, b) ecosystem, c) Swiss army knife, and d) mega-platforms. Importantly, as pointed out before, the typologies are not mutually exclusive and super apps can present a combined model. This finding points back to our initial research question regarding the definition of super apps and their trajectories throughout time. Their evolution and type are closely intertwined to the cultural and geo-political context where they operate and later expand.

Within this fluid typology, we found that the Google Play Store categories Finance, Maps & Navigation, and Communications dominate in the tagging for super apps, suggesting that many of these apps grew from or are primarily used for these functions. While it may be useful to confirm this trend on a larger scale, it provides insight into the types of apps that are primed for ‘super-appiness’. This is also reflected in our methodological tagging of app features, where Payment & Finance, Gig Economy, and Shopping all emerged on top. Payment & finance, a function that both gig economy and shopping both rely on for transferring of services, stands out as integral to an app becoming ‘super.’ It is worth noting that from our geographical analysis, we found that many of the super apps that began as payment & finance apps are located in African countries. This begs consideration of what types of values are instilled in payment & finance apps and how might this inform the growth of a super app.

Conclusions

At its core, this investigation asks, what is a super app? We found that there is no one type of super app, and thus, any future research or regulation regarding apps that might consider themselves of this structure type requires accounting for this ambiguity, as well as local geographic, economic, and cultural dynamics at play. Regulators might also consider whether an app calls itself a super app, and when that might work to their advantage. For instance, while Facebook falls under a number of the categories in our super app typology, it does not call itself a super app. Avoiding this label may enable it to expand in a more discreet way, especially in the context of antitrust regulation in the US.

These findings serve as a launching point not only for regulation, but also determining the benefits and drawbacks of a super app. Further points of research could be: are super apps successful, i.e. are they adopted and then dominant in their particular regions? What super apps have failed, and why? What are the risks of super apps, especially in the age of monopolization? How do users feel about super apps? How about those that are contracted by them? How does region impact the design, development, and trajectory of a super app? For example, would an app that starts as a ridehailing app follow a similar path to super app in Latin America, North America, and Southeast Asia?

This research provides a framework to deal with the ambiguities and complexities of super apps and can as well as should serve as the basis for future research.

Bibliography

Ajene, E. (2021, August 8). The rise of African super apps. Medium. https://medium.com/@eajene/how-africas-super-app-landscape-is-evolving-6c2e5eca6b2f

Chen, Y., Mao, Z., & Qiu, J. L. (2018). Super-sticky WeChat and Chinese society (First edition). Emerald Publishing.

Gamba, J., Rashed, M., Razaghpanah, A., Tapiador, J., & Vallina-Rodriguez, N. (2020). An Analysis of Pre-installed Android Software. 2020 IEEE Symposium on Security and Privacy (SP), 1039–1055. https://doi.org/10.1109/SP40000.2020.00013

Gerlitz, C., Helmond, A., van der Vlist, F. N., & Weltevrede, E. (2019). Regramming the Platform: Infrastructural Relations between Apps and Social Media. Computational Culture, 7. http://computationalculture.net/regramming-the-platform/

Dieter, Michael, et al. ‘Multi-Situated App Studies: Methods and Propositions’. Social Media + Society, vol. 5, no. 2, Apr. 2019, pp. 1–15. SAGE Journals, https://doi.org/10.1177/2056305119846486.

Goggin, G. (2021). Apps: From mobile phones to digital lives. Polity.

Google. (n.d.). Play Console Help. Retrieved 8 July 2022, from https://support.google.com/googleplay/android-developer/answer/9859673?hl=en#zippy=%2Capps

Hoernig, S., & Monteiro, F. (2020). Zero-rating and network effects. Economics Letters, 186, 108813. https://doi.org/10.1016/j.econlet.2019.108813

Nieborg, D. B., & Helmond, A. (2019). The political economy of Facebook’s platformization in the mobile ecosystem: Facebook Messenger as a platform instance. Media, Culture & Society, 41(2), 196–218. https://doi.org/10.1177/0163443718818384

Steinberg, Marc. ‘LINE as Super App: Platformization in East Asia’. Social Media + Society, vol. 6, no. 2, Apr. 2020. SAGE Journals, https://doi.org/10.1177/2056305120933285.

Poster presentation

• Poster 1.pdf: Poster presentation (1 of 3)
• Poster 2.pdf: Poster presentation (2 of 3)
• Poster 3.pdf: Poster presentation (3 of 3)