Revived from 2021 and updated. The word “platform” has only gotten more overused since.
Software vendors rarely describe what they sell as a product anymore. It is a platform. Extensible, open, an ecosystem. But an API and a plug-in architecture do not make a platform, and the difference matters a great deal if you are deciding what to build.
What a platform actually is
A product does a thing. A platform is a thing for things: an intermediary that lets two or more groups of users create value with each other. What makes it a platform is not the technology. It is the business model around it: how you recruit the people who build on it, how they make money, and how you make money from what they build.
Three definitions that hold up:
- “A platform is something that lifts you up and on which others can stand.” Other businesses connect to yours, build on it, and create value together.
- Products are things you can buy, sell, make, or lend. Platforms let one product work with other products.
- Bill Gates’s test: it is a platform when the economic value of everyone who uses it exceeds the value of the company that created it.
Why companies want to make the leap is simple. A product produces one revenue stream. A platform can produce many. Amazon was a retailer before it opened the Marketplace. Google was a search engine before search advertising. Apple sold iPods before the iTunes Store and the App Store turned the device into the front door to an economy.
The three parts
A useful way to think about what a platform needs, borrowed from the people who study them: connection, gravity, and flow.
The toolbox is the technology that lets participants connect: the APIs, the integrations, the applications that show what is possible. Note that Apple and Google publish APIs and still build the key applications themselves, because a platform with no applications attracts nobody.
The magnet is what draws both sides in. Usually a combination of a real innovation and a business model that makes participating worth it. Early eBay matched buyers and sellers cheaply and then added the review system that made strangers trust each other.
The matchmaker is what creates value for the participants once they are there. The App Store is a curated, secure storefront for consumers and a distribution channel for developers at the same time. Both sides get something they could not get alone.
Uber and Lyft did not change the need to get from one place to another, and the technology is not the hard part. They are platforms because they brought a huge supply of rides that were previously inaccessible to a large user base through a frictionless payment and communication layer.
The questions to answer before you try
Most companies that set out to build a platform should build a product first. The path from one to the other, according to the research that has looked at companies that made it, is deceptively simple to describe and hard to do:
- Start with a defensible product and a critical mass of users. Nobody builds on a platform nobody uses.
- Design a business model that creates new value and shares it with the participants.
- Convert your existing users to the platform quickly, before someone else does.
- Watch for and act on the ways competitors could imitate you.
If you cannot yet answer who the two sides are, why each one shows up, and how each one and you make money, you do not have a platform strategy. You have a product with an API, which is fine, and often the right thing to build.
Where this shows up in our work
The question comes up most often in new product development, when a company that knows its industry well sees a chance to become the place where that industry does business. Sometimes that is right. Often the better first move is a product that solves one side’s problem completely, with the architecture left open for the second side later. A VDP Sprint is where we work that out, on paper, before the expensive decisions are made.
A former client once asked us about an “Uber for dog walkers.” The slogan, we suggested, should be “Taking your dog from Point A to Number Two.” The business model question was harder.