Ruslan Leteyski built Checkout X, an alternative checkout for Shopify stores, to around €600,000 a month in recurring revenue. Bootstrapped. Years of work.
Then Shopify locked down checkout access, and it was over. Not declining. Over.
He has written the story himself, in two parts, and the title he gave it is exact: bootstrapping to €600k MRR and getting killed by Shopify. He stayed in the same category afterwards, bought a small app, and co-founded another business that’s now doing several million a year. The recovery is a good story. The kill is the useful one.
Shopify was in excellent health
This is the part that gets skipped, and it’s the whole lesson.
Shopify wasn’t in trouble when it ended Checkout X. It wasn’t cutting costs, cornered, or fighting for survival. It was strong, and it reclaimed a surface it had previously let other people build on, because it was strong enough to want that surface back and to take it without consequence.
Nothing about that is a betrayal. It’s what a healthy platform does. A company with a functioning strategy will eventually look at the most valuable thing happening on its edges and decide to own it.
You’re diligencing the wrong axis
Ask a founder about platform risk, and you’ll get an answer about survival. Are they funded? Are they profitable? Will they be here in three years? Reasonable questions, and they’re aimed at the wrong thing.
There are two axes here and they move independently.
Solvency is whether the vendor continues to exist. Policy is whether the vendor continues to permit the specific thing you’re built on. Solvency risk is what founders check. Policy risk is what shuts businesses down.
And the two are not merely different. They pull against each other. A struggling platform needs its ecosystem – every app, every integration, every reason for a customer to stay – and it isn’t in a position to take anything back. A healthy one, especially one that answers to public shareholders every quarter, has both a reason to reprice the edges and the strength to withstand the complaints.
Good health is the enabling condition. The safest-looking dependency in your stack is the one most able to change its mind.
This is live for anyone building on AI right now
I’ll be concrete, because the general version of this argument is worthless.
Every product being built on a model API today is built on a policy, not a contract. Rate limits, tier pricing, context windows, what a model is allowed to be used for, whether a capability stays in the cheap tier or moves to the expensive one – these are settings, and settings get changed by the people who own them.
The reflex right now is to read a model provider’s commercial strength as reassurance. It is reassurance, on one axis. It’s the opposite on the other. A provider under real earnings pressure has a live incentive to reprice developer economics and the strength to do so.
My own exposure, since I’m asking you to look at yours
I run a validation product on a model API. Its entire output is long-form reasoning – eight documents, tens of thousands of words per venture – and that shape only makes commercial sense at a particular ratio between what a customer pays once and what the tokens cost me to produce.
I don’t control the second number. If per-token pricing on the tier I use changes materially, my one-time price stops working, and I’d be repricing a live product based on a change I had no notice of. The product still runs. The business model is what breaks.
Writing that down took about ten minutes, and it’s the first time I’ve done it plainly, which is roughly the point of this post. It is the same discipline as noticing that the best product doesn’t win – the thing that decides your outcome is often not the thing you have been working on.
The question worth answering this week
Not “should I be on this platform.” You’re already on it, and that question has no move attached to it.
Ask instead: which capabilities of mine exist only because a vendor currently permits them?
Write the list. For each line, answer three things. What does my product still do if this is withdrawn? What does it cost me if the price triples instead? How much notice would I actually get?
Most of the list will come back fine – annoying, survivable, a week of work. One or two lines won’t, and those are the ones worth building an alternative for, or at least worth knowing about before someone else decides for you.
Checkout X’s founder didn’t lose because he picked a weak partner. He lost because he picked a strong one, and strong partners eventually want the good parts back.
Want more like this? Rick writes about the go/no-go decision, founder counterintuitions, and the business of building ventures worth building.
All writing