When we launched https://glideapps.com, we expected to use PWA as a temporary platform as we tested our product theories. The advances Google, Microsoft, and even Apple have made in PWA support since then totally surprised us, and our web-based apps have taken us further than we ever imagined. I've always been skeptical of non-native but I can't argue with what I see every day.
It blows my mind that complex desktop apps-think spreadsheets and image editing--are moving towards the web while every online forum and retailer offers a native mobile app. It's the antithesis of reason.
Complex functionality that doesn't depend on a remote resource belongs in a native app, whether desktop or mobile: Spreadsheet, CAD, word processor, etc. Online shopping, social media, and other content-based services that rely primarily on server-side resources gain nothing from native mobile apps.
If nobody wants to download apps, that's kind of an admission that walled garden is fundamentally flawed. Walled gardens surely benefit their owners, but for the rest of us? I'm unsure.
The value, as I see it, in a native app for an ecommerce company lies in personalized push notifications (transactional and marketing). However, encouraging users to download an app solely for that purpose seems like an impossible task.
I'm of the mind that holistic "let the customer decide" approach is best. Have a very solid strategy for a PWA, native app, and your marketing/notification funnel. As long as the opportunity cost of developing and maintaining an app isn't prohibitive....
The value proposition of downloading the app (in the ecommerce space) only makes sense if I'm regularly purchasing from you. And the only retailer that large might be Amazon. Anything else that's niche, I'm not buying all that frequently.