In the case of the software in question on this thread, the software is useless without the online service.
It's very hard to price a lifetime, one-time-fee license that requires ongoing expenses to provide an online service, in such a way that providing the service in perpetuity is sustainable.
And these people didn't.
Re lifetime licenses, if you offer it for a certain amount of time or until you reach a goal. it can work. At that point, hey! You have funding and you bootstrapped your product/company.
Then you switch to whatever licensing scheme you want.
It would certainly appear these guys did not know how.
As new release dates near, the problem is that the old product has diminished value (or at least is generally perceived to), the new product isn't generating revenues but is generating costs, and the hint of a new release means that the old product has vastly reduced price pressure. The hardware case of this is known as being Osbourned, after Osbourne Computer's ill-timed next-gen product statements.
With subscription service, cashflow is consistent, and is time-averaged over the development/release cycle. From a producer's perspective, this is an upside. For the customer, it's a downside, as risk is transferred, and an inability to make a subscription payment means loss of access to the service.
My sense is that there ought to be a sensible split, where small customers can outright buy product, but large customers (generally, enterprise/government) are on a subscrption basis. The latter have far more consistent budgeting over time, and also comparatively greater negotiation leverage (so that they aren't entirely beholden to the vendor).
Shrinkwrap licensing is actually a very small case of software sales, and is virtually always secondary to subscription, lease, services (contract), advertising, or per-unit revenue models. That last is Microsoft's bread and butter, penultimate, Google, Facebook, and the AdTech-supported Web.
TL;DR: You buy the current release, but there's (possibly) no future.
I’m having trouble thinking of a significant collection of medium to large-sized successful software companies that still operate that way.
Video Game companies are perhaps a decent example of that model sort of working, except that the industry has a history of constant studio dissolution and poor working conditions.
So, it’s not that I’m assuming you’re wrong, I just think your idea needs to be backed up with some more evidence. It’s easy to declare that everything about the floppy disks in a box software model works fine but I’m not sure reality supports the hypothesis.
I give you money, you give me a CD or a disk or a download with the software on it.
Maybe it doesn't work well with perpetually broken, constantly evolving modern software that is never done.
Microsoft didn't rely on shrinkwrap, but on:
1. Per-unit CPU licensing.
2. OEM contracts.
3. Enterprise sales of an "integrated ecosystem" (a/k/a tied/bundled OS, Office suite, and network / directory management).
Shrinkwrap was just gravy.
Most major software firms are actually consulting companies, with a massive services division: IBM, Oracle, SAP. Saleseforce might be an exception.
There are a few traditionally shrinkwrap firms, of which Adobe is probably the most notable. They've been shifting to subscriptions.
Most current Mobile software dev is based on advertising and/or surveillance capitalism.
If you download the app from their website (not the Mac App Store), then there’s an option in the app to buy a standalone license.
I think this points to another factor (besides the cost of running servers) in why so many apps are moving to subscriptions instead of the old purchase-outright-and-pay-for-upgrades model: the iOS and Mac app stores have never supported that model. There’s no way to offer a discount to customers that are upgrading from a previous version.
Apple introduced app bundles a few years ago, and those can now be used, with the same caveat of putting up the new version as its own app. In this case the App Store itself can do the check for the original version and change the price of the bundle.
It's certainly not ideal, but it is possible. Pythonista and GoodNotes are two that I know of that have used each of these workarounds.