Picking Shopify as a landlord

Two years ago I wrote down that every platform I'd ever built on was gone or unrecognizable, and that I kept being surprised by it. This year I started a company that exists entirely inside the Shopify ecosystem.

Those two facts sit next to each other uncomfortably, so let me argue with myself in public.

The case against me#

I know how this goes. MySpace, organic Facebook reach, the ad auctions, the reach algorithms Exposely's supply side depended on. Four times I built something on ground I didn't own and four times the ground moved, and in every case my plan had quietly assumed it wouldn't.

Now I'm on someone else's platform again, with a product that requires their APIs, distributed through their app store, discovered through their search, billed through their billing. If Shopify decides tomorrow that cart recovery is a first-party feature, or changes how apps access checkout, or reprices the store, I find out when everyone else does.

I have no counterargument to any of that. It's all true and I did it anyway.

Why I did it anyway#

The distribution is the product's largest input. A merchant with an abandonment problem searches the app store, finds us, installs in a click, and is billed by Shopify without ever giving us a card. I have started a company with no distribution before. It took three years to build a sales motion that a listing gives me on day one. Refusing that on principle isn't discipline, it's choosing to be small for aesthetic reasons.

The customer is qualified by definition. Everyone on Shopify sells things online, has carts, and abandons them. I don't have to explain the problem, size a market, or educate a category — the entire painful first act of Exposely doesn't exist here. That's worth more than I would have guessed before I'd lived without it.

Alignment, at least for now. Shopify makes money when merchants make money. A product that recovers revenue for merchants is pulling the same direction. That's not a guarantee of anything — alignment is a fact about today, not a promise about next year — but it's a materially different position than being on a platform whose ad business competes with you for the same dollar.

The distinction I'm trying to hold onto#

Here's the thing I think I finally have language for, and it's the reason I'm comfortable this time.

There's a difference between a product that uses a platform and a product that is a behavior of the platform.

LiveRecover sends text messages to people who abandoned a cart. The abandonment data comes from Shopify. The messages don't. If Shopify changed dramatically, we'd have a bad quarter, a migration, and a real amount of pain — and we'd still be a company that texts shoppers, because the actual mechanism lives outside.

A product built into the checkout itself, or one that depends on a specific hook that only exists because Shopify chose to expose it — that's a different exposure entirely. Same "platform risk" label. Not remotely the same risk.

I'd like to say I'll apply this test rigorously forever. My track record on applying my own lessons is four for four in the wrong direction, so let's see.

What I'm going to do differently#

Not much, honestly, and that's the point. I'm not going to refuse the distribution. What I'm going to try to do is hold two things at once: build like this will last, and price the revenue like it might not. Take money off the table earlier than feels necessary. Don't hire ahead of it. Don't let a valuation assume permanence.

That's a much less satisfying conclusion than "own your platform." It's the only one I've found that survives contact with actually wanting to build something.