The issue is, most of the "apps" do not need to be apps, they can well be web pages. Everyone thinks they need "an app" because it's hot this year. But for apps that make more sense being an app, native is much better solution.

For example, for apps that need to store data and work reliably offline, I'd much more trust a native app than any PWA despite of all the services workers and other stuff that it implements.

(Side rant: I hate governments spending public money on apps for two selected, closed ecosystems, instead of producing a good interoperable webapp).

From purely technical point of view, in my experience (writing Cordova app, and reading about NativeScript and other technologies) is that all of the "write once, use on 10 platforms" frameworks are typically more polished for iOS than for Android, or vice versa.

We've built two Cordova apps in the last time, one with Ionic, one earlier with proprietary framework. While they work okayish on Android, the UX on iOS is rather horrible due to a number of smallish issues that when added up, sum to a big difference.

Next, with Cordova et al. you can do easy things easily, but hard things are hard or impossible; you are always behind the native platform capabilities (which change very often and fast), which give a lot more power.

Edit: The point of cost of getting users into a native app is fair, I very very rarely download new apps. This only gets harder as app stores get bombarded with thousands of apps each week. But it doesn't mean native apps should be totally abandoned. They have their place in the ecosystem.

Swift Developer: Native apps are better!

Reality: both will be around for a long time and both have their appropriate use cases. At some point they may converge but we are far from it.

Also, creating any app (web or native) and hoping to achieve widespread success is a gamble.

https://www.quora.com/How-do-I-record-audio-using-the-iPhone...

Remember, due to Apple regulations, any browser on iOS (including the Chrome app) is actually just a thin shell around Safari. It HAS to be Safari, no custom browsers allowed.

This really stifles progress when it comes to the many "bleeding edge" web technologies that are required for web apps to truly replace native apps.

Unfortunately, browsers are still far behind in general.

Development is expensive whichever way you want to slice it. Write Once, Run Everywhere is equally expensive because you have to test everywhere as well so the economics aren't as straightforward as the article suggests. If it was Java would have won years ago.

A counter argument could also be made that a well curated ecosystem of 16% is cheaper to developer for than an ecosystem of 84% made up of a whole mix of different supported levels.

It's really about content and value -- and in a way, discoverability and marketing. Like the article says, most apps are never downloaded. It stands to reason that websites are nearly never visited. Those that are, are found through search results, advertisements, or out-of-band knowledge (e.g. friend recommendation)

Even the examples are not helpful: Alibaba is a market leader. Weather.com is a market leader. I've never heard of Housing.com, but they have one of the best, most valuable, most-obvious domain names in the world. Perhaps they increased conversations and reduced user friction, but these platforms could have only lost business to those of their competitors who clear the bar of 'discovered', 'advertised', and get those user conversions in turn. Quoting their smaller competitors would be far more impactful of the article's point.

The fact is, the app boom is clearly over, top apps dominate user traffic, and it's even harder to become a breakout hit than it was in the past. But native apps still have their uses: seamless homescreen presence, sophisticated layouts (drinking game: drink every time someone brings up flexbox as a counterpoint), offline operation (using mature tech, not serviceworkers), integration into iOS preferences, integration into Android intents, OS-native permission control, etc. These advantages are not negligible over browser-based solutions.

"You may be thinking that PWAs are a non-starter until iOS starts to support the manifest standard and service workers, but I have news for you:

You can achieve PWA-like behavior with Apple’s proprietary meta tags.

There is a service worker polyfill for Cordova that will allow you to build a hybrid app for iOS, so you can get the PWA benefits without sacrificing an app presence on iOS."

Cordova is a JavaScript/HTML based app framework:

https://cordova.apache.org

So using Cordova means you have to submit an app and hence is not pwa right?

Can non native web apps on mobile get access to push notifications, the camera (low level like snap chat), the photo library, audio recording or touch the address book? As of today I don't think you can do these things without an app on iOS.

If you look at the top paid/free apps on iOS most of them are either multimedia heavy (Instagram, snapchat, Pinterest, etc) or games. Neither of these types of apps would work well with the current state of PWA.

That being said having a separate android/iOS team and shipping apps to two app stores is expensive and a total pain. I think that some major changes in the way apps get developed are on the horizon.

I think by looking at react/react native you can see where things are going. I.e business logic and API layer shared cross platform, cross functional teams that can work on 60-70% of your products surface with domain experts dealing with platform specific issues. That plus the ability to push over the air updates is going to change things quite a bit.

I feel this approach is valid even for large parts of nultidmedia heavy apps (sign up flows, pop up dialogs, etc) and could even be used in a hybrid approach with a pwa style App.

The flaw in this argument I think is that does the author think you can build a great PWA for less than $100k either?

In my experience of doing heavy duty mobile web, it is horrendous to polish to a high standard. It was actually more expensive than doing native apps (before even Xamarin/RN) if there was any tricky parts in it.