Same catalogue product sold two ways: the subscription version has furious reviews and the one-time purchase does not

This is the cleanest natural experiment I found while reading review piles, because it is one product type sold under two models on two platforms.

Catalogue products, the kind where the thing you are buying is a library: watch faces, widget packs, icon packs. On one platform the category runs on subscription. Headline rating 4.56 across 22k reviews, so it looks healthy. The low star pile is almost entirely one complaint, and it is not the price:

"Paid for the year, the app has no faces in it, black screen." "Advertises ten thousand, there are about twenty." "An ad showing a face that is not in the app." And the one that names the mechanism: "you have to pay just to browse, how am I supposed to decide before I can see what exists?"

On the other platform the same category is sold as a one-time purchase, and the equivalents sit around 4.5 to 4.7 with nothing like that pile.

My read: for a catalogue product the subscription is not just a payment model, it is a decision to hide the catalogue. Hiding the catalogue is what makes the marketing sound like a lie, and the refund follows the disappointment rather than the price.

So the rule I took away is narrow but I think it holds: let people browse the whole library free, charge at the download. And whatever you put in the ad had better be in the library.