Drives me nuts that even just websites, like Reddit, push so hard to get you to install their app.

I think it is not primarily related to notifications and telemetry, although that's obviously part of it, or to access to your contacts or whatever, like in the case of FB. My theory is that it's much simpler and dumber than that: that "number of app installs" has simply become a metric that managers can use to promote themselves inside a company, and that companies can ultimately use to sell shares. It's an engagement metric, and since there is no metric for "home screen bookmarks made", there is no way for companies and managers to gamify the web site. Therefore, they simply don't care about the website, except as a means to get you to install the app. It's all a bullshit numbers game.

Because web-based apps don't have a good user experience. Even a web app embedded inside of native app feels bad. Maybe it's something about WebKit specifically that feels bad, but there is a very noticeable difference between a web app and a native app. Part of the issue is the thread model. Touch events come in on the app's main thread, and then those touch events get dispatched on to the JavaScript thread, and then the JavaScript thread needs to dispatch to yet another thread to render. It's just impossible to achieve native-level touch interactions in a web-based app.

And then there is there is the polish level. Apple has spent god knows how long polishing each and every native interaction. All the animation curves are perfect. You lose all of that polish when you build a web app.

> Why do companies want an app?

It's not for telemetry (at least at the scale/size of companies I deal with). As far as I can tell it's the prestige. I know this because our initial app had no push notifications and the clients who paid for it wouldn't stop asking about an app and didn't even touch the (identical) web version.

Push notifications are part of it for sure but I just wanted to give an example of a case where there were no features that were app-only but all they cared about was the app. I'd say "Yeah, we have that new feature and you can see it at yourbrand.ourcompany.com" and they'd say "Ok, when will we have an app?", felt like I was taking crazy pills but I've seen this exact same conversation repeated at multiple companies, multiple clients. And lest someone thing this is a subdomain-related thing, it's not, I had the same questions when we setup the web version at theirdomain.com.

AFAIK this is a bunch of the most common reasons, somewhat ordered from most-important to least-important.

1. It puts an icon on your phone's home screen which reminds you it is there and encourages reuse.

2. They can put notification counters on said home-screen icon.

3. It is generally easier to convince people to enable notifications on the app than on the website.

4. User perception. Some users expect an app and look down on a website. (Even if functionally identical)

5. Blocking ads on websites is much easier than blocking them on mobile.

6. It can provide better tracking and analytics.

7. The app can provide a better experience.

7a. The ability to pre-download the application makes it less costly to provide more features and code. The just-in-time downloading of web applications means that the size of code, data and resources generally needs to be more carefully considered. Of course you can do careful code loading for webapps but it requires custom code instead of being managed by the store.

7b. Native apps generally have access to more APIs especially without needing to request special permission (contacts, device sensors, large amounts of storage, etc)

7c. Access to native UI toolkits can make the application fit into the expected UX better.

Apps are easier to use than a browser. The flow of installing an app from the app store, and having an icon on the home screen, makes a lot more sense than opening a browser and navigating to a URL.

Web browsers still have some remnants of being a “user agent”. Apps, by contrast, are agents of their creators, noone else; certainly not the user.