The pains of growing a platform

Main illustration: Becky Simpson

For SaaS companies today, everyone wants to be the next big platform. But if you think it’s as simple as having some integrations and public APIs, think again.

We’ve talked a lot about platforms on our blog, released public APIs and we even run a partner program. Today, there are over 250 integrations built by third parties and nearly 40% of our customers are using our APIs.

But that isn’t enough to call ourselves a true platform just yet. Becoming a platform doesn’t happen overnight, and along the way we’ve encountered many pains and learnings by trying to transform ourselves from a product-first company to a platform. Here’s our story of how we’ve built the Intercom platform.

Platform is probably one of most misused terms within the startup community.

What is a platform anyway?

To understand what a platform is, think of a product like a house. Every house needs to be habitable for the people living inside. They come in all shapes and sizes, serving a variety of needs. But houses need to be connected to the energy grid, have pipes for water, sewage and gas, access to fiber or phone lines and be registered in the municipality. Without these integrations a house wouldn’t be habitable. Well, with the exception of self-sustained buildings ?

If products are houses, then platforms are entire cities. Cities are fully interconnected networks, ensuring all the houses and buildings (i.e. products) within it are well supplied and governed with systems, rules and vision. So think of a platform as a governed ecosystem, rather than just some APIs and integrations.

Everyone wants to build a platform

Nowadays, most SaaS companies are striving to become platforms. By getting people to build on top of them it allows them to grow even faster, no longer constrained by their own engineering headcount to progress the product. This drives new sources of revenue for the company, extends existing products to new use cases and crucially, makes products stickier to customers. Once deeply integrated with consolidated data records, it becomes dramatically harder to switch to a competitor.

While platform plays are a dime a dozen, making the steps towards becoming one aren’t so straightforward. Some companies like Salesforce, Box and Slack have managed to successfully become platforms, but not without a cost. For example, Slack had to seed its marketplace with integrations of their own. Today, they have a rich ecosystem of integrations that helped customers integrate Slack deep within their business workflows, but it was a result of deliberate planning rather than just “opening up”.

Building a lasting platform is actually quite complex but is made infinitely simpler by avoiding some of the following mistakes.

Mistake #1: Trying too hard, too early

If you focus on giving value to customers, platform will naturally follow.

“Platform” is probably one of the most misused terms within the startup community. Everyone loves using it as it makes a product sound like an established company. In fact, before I joined Intercom I recall describing my own startup as a “platform of X” when pitching our product. Platform embraced a bigger vision than a simple product, so it was easier to pitch it to potential investors.

It’s also a misused term in growing businesses. It’s an easy pitfall for companies to think of platform as the next milestone to achieve once they have found product-market fit. At Intercom, we’re trying hard to avoid using the term “platform” during product discussions, so that we can be more precise with what we truly want to achieve. If you focus on giving value to customers, platform will naturally follow.

Building a platform takes years of progression. In his book Marc Benioff explains how this happened with Salesforce. Small steps like allowing customers to rename tabs led to a cascade of changes that ended up with creating custom fields, objects and workflows. Salesforce moved from a point solution (i.e. solving one particular problem very well) to a platform. This followed with continuous investments like opening up Force.com and acquiring Data.com.

What we often miss is that platforms are extensible by nature. It’s not just about sending data in and out of your product but providing an infrastructure for others to build and grow their business by accessing your underlying data. Unlike products which serve one or just a few jobs, platforms are an ecosystem of endless jobs served through flagship products, third-party products or customers building for themselves. Customization, for better or for worse, is a crucial attribute for platforms.

Takeaway: Before you start talking about platform, think of how deeply extensible your product is.

Mistake #2: Others will build the integrations

At some point, most SaaS companies start to release public APIs. They realize you can’t make all the integrations customers are asking for and hope that exposing some APIs will fill the gap. This is when they start believing in bad dogmas like “others will build it for us”. In reality, it rarely works like this.

What makes businesses build on you is distribution.

Strategic integrations are crucial for the success of your product no matter your size and you should always expect to be building integrations yourself. For some of them, you want to fully own the experience and make sure they are deeply integrated. This lets you dogfood your own APIs and find the right gaps that need roadmap attention. More importantly, you want to understand deeply how your product fits in your customers’ workflows, rather than hand off to a third party.

A good example of this was our recent Salesforce integration. It was a blocker for many of our customers so we prioritized the CRM use cases we wanted to focus on and progressed the platform to serve these needs. In other words, we exposed impactful APIs, created abstracted services and explored new experience components in our interfaces. Building platform for the sake of platform would have got us nowhere.

Takeaway: Before building a platform for third parties to sell their services, start with one for your customers to build on.

Mistake #3: Focusing on platform adoption, rather than developer success

How do you measure the success of a platform? One way to evaluate platform health is the amount of integrations available. In the early days, this count can be valuable for potential customers who are deciding to purchase your product. But it can also seriously mislead your platform teams. Even by taking customers into the equation, such as the average number of integrations per customer or percentage of “integrated” customers, you’re still only seeing one side of the picture. Platform metrics should have developers squarely in mind.

We’ve started to think deeply about what a preferred integration looks like. It’s not about disqualifying any original (yet valuable) use of the platform. But there are certain areas that might be a strategic focus for the company. This is tightly connected to shaping a strategy around your platform and giving it a clear purpose beyond “opening up”. The question to ask yourself is – how can we drive the right type of integrations?

If we shift the table and start thinking about how we can get developers to succeed on our platform, we will implicitly make our platform valuable to our customers. However, to attract the right developers you need to have platform features they consider valuable and can help them get exposure. At the end of the day, it’s a business and they are looking to grow their business with you. Defining developer success early on will help you prioritize efforts.

Last but not least, you need to also actively invest resources in engaging, motivating and educating developers in order to get them building with you. What makes businesses build on you is distribution. It doesn’t mean you need a marketplace right away. If you are looking for third-party developers to join your platform then you need to think how you can make them succeed for the long run, not just getting them started.

Takeaway: Before measuring platform metrics, start by defining what developer success looks like.

Building platform takes time

There are no silver bullets when it comes to building a platform. A sticky and healthy platform takes time, with none of the immediate feedback loops you get with your other product features. When you get your platform’s features into your customer hands, you need developers not just to adopt it but to sell it to your customers too. Building a platform doesn’t happen overnight, so stop hoping for immediate success and start building for long-term value instead.