Browsers / web apps are a more portable, and more productive method of distributing GUIs. Full stop.
When deciding to go native or not, you have to decide: Will you hire several designated IOS application developers, several Android application developers, and have them re-implement your website? Or will you hire the equivalent amount of front-end engineers and have them collaborate and develop the codebase together using a shared toolkit?
People moan about how it's slow, but they tend to miss the point: I don't work for a "writing native apps and make the apps fast" company. I work for a software company that builds a product and sells it to users. Developer familiarity == Speed == Money
And really: Do you honestly believe spending 300k a year on engineers to re-implement the same shit we've already written, but in a different language with completely different semantics is a good investment? Rewriting all of the work we've ALREADY done for our mobile view in CSS and HTML and Javascript in Objective-C and Java?
Unless you're writing software with INTENSE graphical requirements, and even THEN there is webgl, writing a native app brings no value. You could spend that time optimizing your website written in JS you've packaged as an IOS/Android app. Or, I don't know, implementing new features and actually bringing value to your company.
PWAs are much slower, but as time goes on this will be improved.
As for the electron haters: Show me a more productive way to implement GUI applications. Show me the applications and frameworks that are putting out GUIs that look even remotely as good as the Electron applications.
I think we're going to see a rise of PWAs and apps that are simply web browsers in disguise.
Edit: I'll also add that the "not always online" or "bandwidth" argument is bunk. Crazy idea: Package your web application as an app, and include it's dependencies and everything it needs to run. Once you do that, the "download once" advantage of a native app is bunk.