Mathias Meyer
Mathias Meyer


For the first 15 months of Travis CI’s existence as a paid product, customer support went a little like this:

  • New email from customer.
  • Open a console on Heroku to fetch the relevant data.
  • Respond to email.
  • Repeat upon next email.

We didn’t have any support tools, if you can imagine that. Nothing that gave us quick access to the data we needed to look into and solve a customer’s issue.

But you know what?

It was okay. It was okay to fake it until we had the time to sit down and make it.

We could’ve spent the time building these tools much earlier on, no doubt. But with an unproven product it was hard to justify the time when we weren’t fully sure yet if our product would help us build a sustainable business.

Instead we chose to spend the time improving our product to the point where we had enough happy customers that we could pay the bills and had a good feeling that the business would work out.

Of course we also pushed back on working on these tools over time for other reasons, but eventually, we sat down and invested the time to build better support tools.

In the end, that time was well worth it, as better support tools that everyone could use meant that less technical people could jump in on support and learn more about our customers’ problems and how Travis CI works.

In the end, the effort benefited everyone. But initially, there was a trade-off between spending time on product to prove its viability vs. spending time on tooling around a product you’re not certain yet whether it’ll take off.