Innovation in Face of Uncertainty

Laurent Paris

SVP Engineering at Datadog



One of our strategic goals at my previous job was to develop music recommendations. No one really did it before and we sought inspiration in similar projects like news and movie recommendations. There was, though, some rudimentary music recommendation but not in the streaming context and streaming introduces a great many variables like when the song was played or from which device. We manage to collect a lot of very-fine grained data. However, I knew from my past experience that in these situations there is a high level of uncertainty and that you have to try as many things possible. To do so we had to create a culture of innovation across the organization.

Actions taken

To create a culture of innovation we had to apply a trial and error approach trying as many things possible. We were not after a concrete solution but after a structural setting that will encourage innovation. We created a system of continual feedback where everything we did was recorded. Therefore, we could assess how much something that actually worked contributed to the overall goal. Having in place such a structural setting was a strong incentive.

We didn’t know in advance what will work and what not and we had to try as many things possible. This approach required us to lower the cost of experiments -- if an experiment is a low-cost one you can do 20 of them, while if it’s costly you can do only a few. Hence we invested in the infrastructure for experiments automating as much as we could. Eventually, we could run an experiment with an outcome available in a few day’s time.

We created an environment that was encouraging people to come up with ideas -- not only leaders or product managers but we particularly encouraged bottom-up innovations. However, the mere encouragement was not enough -- we had to carve out time for people to work on their ideas. We introduce hack days and every Friday afternoon they could work on their projects and try things out. Twice a year we had a company-wide hack week where everyone could create their own team and experiment with their idea.

Lessons learned

  • Many successful innovations came up from a bottom-up innovation. For example, Discover Weekly was never part of any product roadmap. It was created by an engineer who was passionate about the idea, worked on it during hack days and manage to convince others to join and work with him.
  • There were teams that were tremendously passionate about their projects and for them, this went beyond work. Whether they were passionate about music or the problem itself, the best innovation came from most passionate people. Great ideas could come from anywhere. This is the reason we introduced hack days and supported bottom-up innovation. Not knowing where good ideas would come from we created a culture where innovation can seamlessly flourish.
  • Be prepared to be surprised! What works and what not is very counter-intuitive. Things we were convinced would work -- didn’t, while some that were hesitant about but were cheap to be tested -- worked the best at the end. You have to try as many things as possible. Keep track comprehensively and have solid statistics; numbers will eventually beat uncertainty. You have to balance strong opinions stemming from the passion with scientific, factual verification that relies on collecting and analyzing data.

Be notified about next articles from Laurent Paris

Laurent Paris

SVP Engineering at Datadog

Culture DevelopmentFeedback Techniques

Connect and Learn with the Best Eng Leaders

We will send you a weekly newsletter with new mentors, circles, peer groups, content, webinars,bounties and free events.


HomeCircles1-on-1 MentorshipBounties

© 2024 Plato. All rights reserved

LoginSign up