The app store is a search engine

A year into distributing through the Shopify App Store, the single most useful reframe I've had is this: it isn't a store, it's a search engine.

That sounds like a distinction without a difference. It isn't. A store implies browsing — customers wander in, look around, discover things. A search engine implies intent — someone types words describing a problem and receives a ranked list, and the ranking determines almost everything.

Merchants don't browse for apps. They have a problem, they search the words for that problem, and they install something from the top of the results. Where you sit in that list is not a marketing input. It's the business.

What follows from taking this seriously#

Your listing is a ranked document, not a brochure. Most developers write their listing as a pitch — clever positioning, brand voice, aspirational language. That's optimizing for the person who already found you. The much larger population never sees it because you didn't rank for the words they used.

The words merchants use are not the words you use. We call it cart recovery. A significant number of merchants search for "abandoned cart," or describe the outcome rather than the mechanism. Every gap between your vocabulary and theirs is traffic you never see and never know you missed.

Reviews are ranking input and social proof simultaneously. Which means a review is worth roughly double what it looks like — it lifts you in the results and converts the people the lift brings. Nothing else in the listing has that property.

The category page is a mixed blessing. Being categorized correctly gets you into the browse surface. It also puts you in a row of direct competitors, side by side, where the merchant is comparing on price and review count with no context. I'd rather rank for a specific problem than be one of twelve tiles.

The thing this makes me uncomfortable about#

If the ranking is the business, then the platform decides how big your company is, and it can change that with a ranking change nobody announces.

I wrote last year about the difference between using a platform and being a behavior of a platform, and I felt reasonably good about which side we were on — we send text messages, the messages don't depend on Shopify's internals. That's still true about the product.

It is not true about the distribution. Our acquisition is entirely a function of an algorithm we can't see, run by a company that doesn't owe us anything, over which we have exactly the influence of a website trying to rank on Google in 2005.

So there are two separate risks here and I'd been treating them as one. Product risk: would we survive an API change. Distribution risk: would we survive a ranking change. We're well positioned on the first and completely exposed on the second, and the second is the one that would actually kill the growth.

What I'm doing about it#

Nothing dramatic, because there isn't a dramatic option that doesn't involve giving up the best distribution I've ever had.

What I am doing is building things that aren't ranked by anyone else. Merchants who found us through the listing and were happy tell other merchants — that channel is slow, unmeasurable, and mine. Content that ranks on the actual web for the same problem, so there's a path to us that Shopify doesn't mediate. A reason for a merchant to know our name before they search.

None of that replaces the listing. All of it is a hedge against the day the listing stops working the way it does now, which I'd put at a certainty on a long enough timeline.

The uncomfortable version: I'm running a company whose growth rate is set by a search algorithm I can observe and not inspect. That's the same position I was in as a media buyer, and I said at the time it was a bad long-term seat for a professional. It's a bad long-term seat for a company too.