Why did PWAs fail?

Short answer

They did not fail — the promise that they would replace app stores did, and for two reasons that have nothing to do with the technology. Apple never shipped an install prompt, so on iPhone the user must know a hidden menu path. And nobody solved discovery: an app store is a place people already go, a web app has to be found. What the technology actually delivers, we measured: 151 of 386 installable apps keep working with the network off. The format works where someone did the work.

The iPhone install path nobody completes

On Android, Chrome offers to install. On iOS, there is no prompt and never has been: Share, scroll past the sharing options, tap Add to Home Screen. Anyone who has walked a non-technical person through that sequence knows how it goes. Someone who has onboarded people in live calls put it at roughly one in ten knowing the path unaided.

That is not a manifest problem and no amount of engineering fixes it. It is a platform decision, and it removes the frictionless install that the entire pitch rested on for half the mobile market.

Mentioned here: ✏️ Excalidraw, 🖼️ Photopea, ✏️ Penpot

Discovery, which nobody replaced

An app store is a destination. People open it intending to find software. The web has no equivalent surface for applications — search returns pages, not apps, and there is no shelf to be on.

This is the harder failure because it is not technical at all. The format removed the gatekeeper and, with it, the distribution the gatekeeper provided. Most PWA advocacy treated that as pure gain.

The failure that is real, and self-inflicted

Here is the part the discourse misses, and it is the one we can measure. 235 of the 386 installable apps we tested install correctly and then show nothing once the network drops — almost always the same four causes. The defining feature of the format is absent from 61% of the apps that advertise it.

Manifests tell the same story: of 373 we could read, 329 have no id — meaning one URL change silently orphans every installed user — and 215 have no maskable icon, so the app looks unfinished on the home screen. 3 apps declare a manifest that no longer responds at all.

The fuller picture is in the measurements. So when someone says PWAs feel worse than native apps, they are frequently right, and not because of a platform limit. They are describing apps where the work was not done — and the tooling never told anyone, because none of these defects produce an error.

What survived

The economics, unchanged and still compelling: no 30% commission, no review queue, no gatekeeper who can remove you, one codebase across every platform. For tools and utilities that is often decisive, and the apps that did the work are genuinely good software.

The honest framing is narrower than the 2018 pitch and more useful. PWAs are an excellent way to ship a tool without a gatekeeper. They are a poor way to be discovered, and a poor way to reach iPhone users who do not already know how to install one.

Frequently asked

Are PWAs dead?

No. 151 of the 386 installable apps we measured work offline, install in one click on Android and desktop, and cost their makers no commission. What died is the claim that they would displace app stores — that needed frictionless installation everywhere and a discovery surface, and neither arrived.

Is Apple to blame?

Partly, and specifically for the missing install prompt on iOS, which no developer can work around. But 61% of the installable apps we measured do not work offline on any platform, which is not Apple's doing. Both things are true at once.

Also published on Medium. This page is the original.