About That 85%

And there you have it: in the end it is not the revenue model that makes a developer money. It is the quality of the app and the support the user receives that makes him a returning customer. It almost looks like being able to write proper software has little to do with the success of and app. Being able to properly support an app and properly communicating about it and being receptive to what customers want to do with the application is more important. Likewise is playing the long game is what makes you money.

I sell a simple utlity for the Mac, on both the App Store and from my website. Of those who pay me directly, quite a few gladly, and voluntarily, without me begging them, pay me again after a year or in longer intervals. Because I make a product that people are generally very happy with, and for which I provide good support (meaning I practically NEVER get complaints, and even for the few where people ask for their money back, I can help them figure out how to better use it, and they withdraw their refund request).

And I'm sure that if I'd directly ask people to pay again after a year or two, many would be happy to. No need for subscription. I keep it voluntary, and that's what users appreciate - and as they see that they keep getting good value from the app, they're happy to pay again.

FAANG have agency in this era, and smaller developers do not. And customers? Well, they don't enter into the picture at all.

Not all software can be or should be subscription. For Example Adobe is a great case for subscription, as most business wants the consistent update, support and monthly payment rather than one time payment ( which is easier to reason as a business expenses )

For other software that are more utility like, for example like Soulver, or other smaller tools, with a niche market and little business / enterprise usage. A one off payment with limited update still fits better in most people's mental model. And people will gladly pay for an upgrade to newer version if they love the product. as @Thomas has mentioned above.

Does a new CPU architecture count as "an upgrade"? What about all the APIs that get deprecated, or modified? I agree that "pay for an upgrade if you want" was a great model in 1991 but I'm not sure how feasible it is any more.

I've got piles of old Mac software I bought that I can't run any more, not because there's anything wrong with it, but simply because the system changed out from under it. Most of it is no longer maintained so I couldn't re-buy it even if I tried.

To me this is actually an argument against offering subscriptions; if 80% of monthly subscribers quit after one year then you’re better off charging them up front, they weren’t going to produce long-term ongoing revenue either way.

I’m guessing this logic is predisposed on the idea that a one-off would be more expensive than an annual fee?

Suppose your app is $10. If 80% do quit, they will have paid $10. But the remaining 20% will have paid at least $20. That means your non-subscription app would need to be at least $12 for the same revenue.