The day we doubled sign ups at Intercom

“What the f***!?

Three words you probably never want to hear from your CEO and Head of Growth at once. Except in this case, the chart was the one below and it showed new accounts created for Intercom 🙂

As is the case with most big wins, the insight is obvious in hindsight. It starts here - in early 2014 someone could only use Intercom by passing their user data through API. We could tell from our data that this was a big dropping off point, and as such we spent months shipping experiments to make it easier to connect to our APIs.

In running these experiments, there was a simple but rigorous process we followed, inspired by folks like Ray Ko and Andy Artz who’d done this at Facebook, OMGPop, and a variety of other early stage companies. It looked like this:

1. Establish a goal metric.
2. Identify, instrument, and measure every action that a user could take within that goal.
3. Prioritize which actions you believe most influence the goal metric AND which you can most meaningfully impact. This involves a combination of intuition and analysis.
4. Develop and prioritize a list of experiments to move those actions.
5. Methodically experiment, using your goal metric to determine success.
6. Check in daily/weekly.

Through this process we shipped experiments simplifying the javascript, reordering the signup flow, building out ‘invite a developer’ flows, etc. Yet the critical insight came through a relatively simple analysis - one like we’d run countless times before. However, this time we made an important distinction between a customer connecting to our API and having user data in the product. It turned out that there were many customers that had established an API connection locally but had actually failed to deploy. In other words they hadn’t actually passed us any user data.

In addition to the obvious question of why customers weren’t deploying something they had successfully installed, this analysis also had us look at this step with a new perspective. For the first time, having users in the product became a distinctly different action from connecting to our API. And this then prompted us to ask: what other ways are there for us to help customers get their user data into Intercom?

From there, CSV import was born which ultimately allowed Intercom customers to upload their users without touching our API. Installs doubled overnight. “What the f***!?” is right… who would have guessed that a simple feature like that could have such a meaningful impact on our business. Yet this particular insight isn’t necessarily reproducible to any other business, so instead I like to focus on the process that got us there.

Folks often tell you that growth is about experimenting and stacking up small wins over time - in baseball terms they’ll say “focus on singles.” I do believe that to be mostly correct - most of the wins at Intercom stacked that way. Yet home runs like the CSV importer are possible, albeit rare! For us, the most important lesson was to trust the process we’d built, continue to revisit our assumptions and old analyses, and to hammer away.

In the depths of grinding work like this, I love coming back to this quote from Jacob Riis. “When nothing seems to help, I go back and look at a stonecutter hammering away at his rock perhaps a hundred times without as much as a crack showing in it. Yet at the hundred and first blow it will split in two, and I know it was not that blow that did it - but all that had gone before.”