All about subscription #1
A few years ago, I watched a few episodes of a show called Upload. I never finished it, but the premise has stayed with me ever since.
It's a comedy about a man who dies and gets uploaded into a digital afterlife: a resort called Lakeview, run by a tech company. He lives there because his girlfriend pays the bill. When the money stops, he stops.
The detail I still remember is the basement. There's a floor for people on the cheap plan, capped at two gigabytes of data a month. When they run out, they freeze mid-sentence and stand there in the hallway until the next billing cycle starts.
It's funny that I recognized every part of it.
A capped free tier. A premium tier with a better view. Add-ons priced by the minute. Support agents who live and die by their star ratings. And the person paying is not the person using, which is a problem every subscription team I know has had to solve.
The writers didn't invent a future business model. They took the oldest one we have and pointed it at the last thing anyone assumed was free.
To be fair to the show: it is satire, and its target is real. But its target is not recurring billing. It's a company that owns the only door and charges rent on being alive. Take away the captivity and there's no joke left.
And satire needs a structure the audience already accepts. The scene only works because a viewer can watch a fictional afterlife send a monthly invoice and think, yes, obviously, that's how it would work. Nobody builds a joke on a structure that's about to disappear.
Satire is a durability test. This one passed.
Four hundred years, one structure
In the early 1600s, a printer in Strasbourg started selling a weekly newspaper to people who paid in advance. By the middle of that century, fire and life insurance worked the same way: pay a little, regularly, stay covered.
In the 1700s, books were published by subscription. Subscribers paid half up front, and their names were printed in the front of the book, ordered by social rank. That is a premium tier with a status badge, three hundred years before anyone made a pricing deck.
Then telephone line rental. Then cable. Then the gym. Then discs in a red envelope, then streaming, then Prime, then Adobe, then Spotify, then the AI assistant most of us now pay for monthly without thinking about it.
Same structure every single time. A recurring payment in exchange for continued access.
Four centuries. Nobody has replaced it. Nobody has even improved it much.
What changed is the noun. News. Protection from fire. A dial tone. Channels. Discs. A library. A machine that thinks. Every generation ends up paying monthly for something the previous generation would have called ridiculous.
A sale asks the customer one question, one time; a subscription asks the same question every month. The model is just the question. Your product is the answer. The question has never needed to change.
The debate that keeps coming back
Should we still be a subscription business? Should we change the model?
It sounds like a strategy question. It almost never is. Nine times out of ten there's a value problem underneath it, and the model debate is what we do instead of looking at it.
The current version of it is edge computing. Storage and compute move onto the device, the customer owns the hardware, so the monthly fee should go away. I wrote a whole post about that one and won't repeat it here — short version: where the work runs is an engineering decision, how you charge is a business decision, and people keep welding the two together because for a decade the cloud was the recurring cost. (https://shalinwu.com/blog/moving-compute-to-the-edge-doesn-t-mean-moving-the-business-model-backwards)
What I want to add is that this is not a new argument. It's the same argument with new hardware attached.
Free MP3s were going to end paying for music. What actually happened is that ownership got replaced by a subscription, and the industry makes more money now than it did selling discs. The free ad-supported web was going to end paying for news. Now almost every publication worth reading has a paywall.
Every time the delivery mechanism moved, someone announced a funeral. Every time, the noun moved and the structure stayed exactly where it was.
Local storage doesn't remove the subscription. It removes storage from the list of things people renew for. That's not the end of the model. That's Tuesday.
The part I should be honest about
Here's where the argument gets uncomfortable, and I don't want to skip it.
If storage is the only reason a group of customers pays you, and storage becomes free and local, that revenue leaves. Not "gets replaced by something better" — leaves. That's real money and pretending otherwise is how you get blindsided.
But I'd argue that revenue was never durable to begin with. It depended on withholding something that got cheaper every single year. You can hold a price on scarcity only as long as the scarcity holds, and storage stopped being scarce a long time ago.
Renewal has to sit on something that keeps improving. The model that got better last month. The detection that stopped calling a shadow an intruder. The patch that closed a hole nobody told you about. Those things don't get cheaper over time. They get better, which is a much stronger place to stand.
I could be wrong about the edges of this. There are probably products where the value genuinely is one-and-done — the box does one thing forever and never needs to change — and a subscription there is a tax with no service behind it. I believe that for some things. I don't believe it for anything with a model inside it that we intend to keep improving.
What I'd rather spend the energy on
Three questions. They're harder than the model question, because you can't answer them in a slide.
What is the customer actually renewing? Say it in one sentence, in their words, not your SKU name. If the sentence is a feature list, you don't have an answer yet.
Does the value show up between the charges? Signup value is easy. Day-40 value is the whole business. A subscription that only performs on the day you buy it is a product with a bad payment plan.
What happens on the day it breaks? Every subscription has a moment where the promise visibly fails — the device drops offline, the alert doesn't come, the answer is wrong. That moment is your real churn model. Not the price page.
Answer those three honestly and the business model debate dissolves, because there is nothing left to argue about.
The dark joke in the show
Lakeview is a subscription business with a customer who can never walk away. That's what makes it funny, and that's what makes it horrifying.
Real customers leave every thirty days if you stop being worth it. That isn't a weakness of the model. It's the best forcing function a business can have — a standing test of whether the value is still flowing, graded by the only people whose opinion counts.
The model is not the thing to question. It has outlived every executive who ever questioned it, and it will outlive us too.
Go find out what your customers are actually renewing for, and go make that thing better.
As always, I'd like to know where this breaks. Especially from anyone who has shipped a product where the recurring value genuinely ran out.