Deloitte Digital shipped its first progressive web app in 2018 – the Human Capital Trends app – and as a product manager it was exciting to observe whether the benefits would materialize as advertised.
First, a quick refresher on what constitutes a progressive web app. You’ve probably heard the term “responsive design” before, referring to a website that scales to any screen size. A progressive web app (PWA) can be added to the home screen and opened full-screen on a mobile phone like a native app (i.e., an app downloaded from the app store). Since this paradigm emerged in 2015, commentators have predicted the eventual demise of native apps.
When Deloitte’s Human Capital practice asked Deloitte Digital to create a standalone, immersive distribution vehicle for their annual research about the trends shaping the world of work, we considered native apps, hybrid apps and a progressive web app. We selected a PWA to reach the devices readers were likely to use – work phones and work computers. We also knew that users overwhelmingly don’t download new apps, yet they visit lots of new websites, and we wanted to avoid obstacles wherever possible.
Sure enough, HC Trends app traffic has been evenly split between web and mobile, validating the decision to support more than just mobile. I am confident that mobile apps alone would have reached significantly fewer users. Session length on desktop is roughly twice that of mobile, indicating that users engage more deeply on desktop while consuming bite-sized content on mobile.
Another goal was to foster ongoing interaction with push notifications. This was a qualified success. We observed click-through rates as high as 25 percent - far higher than most benchmarks - but they reached a mere fraction of users because web push notifications work only on certain browsers and just 5 percent of eligible users opted in. In a future release we would like to remind opted-out users that they are not receiving the latest updates.
Finally, we sought to improve performance by eschewing on-demand page load in favor of a background sync where new content was automatically downloaded on the user’s behalf, then loaded from memory when the user visited a page. However, we found it difficult to teach browsers when to fetch new content instead of loading a page from memory, leading to frantic debugging as it became clear users were sometimes seeing stale pages. We resolved the issue by having the browser initially load saved content, then fetch new content in the background for subsequent visits – a workable but not perfect solution.
The big question is: are PWAs ready for the enterprise? My answer is a nuanced one.
On one hand, building a given feature with a PWA takes less effort, on average, than with a native app due to the overhead required to deploy a native app to a device. Web is now capable of device integrations that used to be the exclusive purview of native apps, including push notifications, offline storage, geolocation and camera controls. Many of these are found in the HC Trends app.
On the other hand, getting those core use cases to work seamlessly is hugely variable with a PWA. Not only does it require complex programming with components neither web developers nor native developers typically work with, such as service workers, but these components behave differently on every browser. We found the QA effort for advanced functionality to be more extensive with a PWA than with a native app.
As developers and product teams ascend the PWA learning curve, these activities will become more predictable. We experienced firsthand the challenges that early adopters face when one browser enhanced its support for service workers a few days before the HC Trends app launch. In theory, the change was purely additive – but we observed that it caused unexpected behavior and only partially worked as intended. With little time left, we had to deprecate functionality for that browser.
It is also frustrating that the “add to home screen” function cannot be triggered programmatically in a PWA. That is, rather than a button within the app that calls a browser API, the user must navigate two or three menu steps in their browser. The HC Trends App tries to overcome this by showing an “add to home screen” tutorial to new users, but it’s not a perfect solution.
Eventually, a set of UX best practices for PWAs will emerge to govern adding to home screen, enabling/managing push notifications, and leveraging web functions like social sharing. There isn’t yet a wealth of community knowledge for PWA developers and product teams to fall back on.
In spite of the glitches, I am excited to watch progressive web apps evolve because I believe they are a winning proposition for users: fewer hurdles in order to get an app, faster page load and less storage requirements. PWAs also come in many shapes and sizes. Shortly after HC Trends, Deloitte Digital built an interactive simulation for the Davos World Economic Forum that had a home screen shortcut and opened full-screen, although it did not have bells and whistles like offline mode or push notifications.
One of the key adoption measures for PWAs will be user acceptance. A few major brands have released experimental PWAs but they show no signs of ramping down their native apps, which, in addition to their large install bases, sustain huge IT and marketing organizations. Another sign of the road yet traveled is that although the HC Trends app was marketed as a PWA requiring no download, some users still asked why they couldn’t find it in the app store.
Andrew Pollen, manager, Deloitte Consulting LLP, is an engagement manager in Deloitte Digital’s Seattle studio. Working with visual designers and software developers, he builds custom applications for Deloitte and its clients, most recently the Human Capital Trends app.