Hackathon Culture: Insights from the Display team
At trivago we’re quite passionate about hackathons. Previously, we used to only hold annual external ones (last year in collaboration with Techfugees), but from November 2017 we also promote internal hackathons within the scope of a certain area (e.g. Marketing or Hotel Search), in order to cut complexity and ensure that we stay innovative despite the huge growth of the company.
Considering this incentive to work independently and have more ownership, it was easy to forecast that single departments and even single teams could come up with ideas for their own hackathon.
Within the Content department, the Display team have recently held their own internal hackathon. The overarching goal of this cross-functional team is to describe a hotel and its surroundings through visual and textual content, so users can make an informed decision. The team came up with 5 different project ideas and transformed them into MVPs (Minimal Valuable Products), in order to enhance the quality of all images and reviews. Considering that these two factors are the most important information for the user (after price) when they look to book a hotel, it is a quite an impressive result.
We talked to Pascal Cremer, Product & Engineering Lead for Visual Content, who has been co-facilitating this hackathon pilot for his domain, to share some knowledge and insights about the process and overall project.
Life at trivago: Hey Pascal! Thanks for the meeting, we’re here to talk about the Display Hackathon! First of all, how did you come up with the idea of organizing a hackathon in your team?
Pascal Cremer: The idea was to use the hackathon format to give people the opportunity to explore topics which are able to bring a concrete user value but that are sometimes pushed a bit aside by other more urgent priorities that require everyday attention.
Lat: That sounds like a nice way to try out new things while motivating people. Where there some requirements to select the projects for the hackathon?
PC: The whole team was able to brainstorm, discuss and present their own ideas freely. However, in order to get the most out of this hackathon, we asked our participants to focus on initiatives to enhance and optimize our display content and all related processes in terms of quality and coverage using the latest and greatest technologies within our AWS (Amazon Web Services) cloud computing stack. Furthermore, we encouraged participants to try out new topics like Machine Learning and Deep Neural Networks, to give them the opportunity to see how those would play out in an extremely practical context. Last, but not least, the outcome of each project should have been a presentation to the whole department in order to share the knowledge gained.
Lat: In this regard, could you tell us a bit more about the ideas that made it to the end?
PC: At the end of the hackathon 5 projects have been completed, one for each idea. Without going too much into details, all of them were able to answer the question “How can we enhance the user experience?” in different ways. On images, we applied sophisticated Machine Learning (ML) algorithms to enhance low resolutions to so-called “super resolutions”, and we worked on a scoring mechanism to ensure that our users will only get shown the most relevant and valuable images. We also did some great research on how to detect “corrupted” (broken or invalid) uploaded images, and, on the textual side, we used ML to optimize hotel descriptions, and built a prototype to combine both visual and textual aspects of a hotel in a very smart and innovative way.
Lat: Wow, they all sound really great! And how these ideas were actually brought to life and become MVPs?
PC: That’s an interesting question because it was all quite different from the standard hackathon format. I took part in several hackathons before and usually they’re always taking place in different locations from the everyday work, they give the chance to work with people that you don’t normally work with, they’re generally structured to have daily stand-ups and retrospectives and they don’t last more than two or three days. For this particular hackathon, we decided to give full freedom and ownership to each group (one per idea, each of them with two or three people) and give them five full days. This extended time allowed to focus also on the state of art of the different topics and research in order to learn more and apply the learning in a clever way.
As for the organization, most of them decided to set daily milestones and reporting them to share with the group, however, the approach was highly personal.
Of course, it’s always nice to work with different people and be in a different place, but it was good to shift completely the focus of people that are used to work together to have a different outcome, and I’d say that all of the projects achieved great results regardless of the organization used.
Lat: Seems that despite the different format everything run quite smoothly! Did you face any challenge in these five days?
PC: Well, the five days themselves were obviously quite challenging in terms of time management but we knew it in advance and we made it anyway!
However, the biggest challenge was actually to come up with ideas that were focused on the back-end without impacting on the front-end to ensure to minimize dependencies for the MVPs that can be turned into more extended projects easily. On top of that, I was really proud of people that really challenged themselves proposing ideas that were out of their everyday comfort zone and pushing to apply in practice the knowledge that was mainly theoretical before this hackathon.
Lat: Well, it really seems the perfect trivago story with entrepreneurial passion and fanatic learning. What was the main take away?
PC: It’s actually true and the response to the initiative from the people involved and from the rest of the department was really positive. People in trivago love to participate in this kind of things. Well, the main take away was that we have incredible people in the team that I will definitely like to thank for all the hard and amazing work of these 5 days! Letting these people focus on what they’re passionate about and on what is able to bring the highest user value giving them the opportunity to organize themselves in the way they preferred. This is already applied to the team and also generally in the company, giving people more ownership on their topic, and helping them when needed but trying not to impose ideas from the top.
However, there’s always room for improvement and promoting hackathons as a possible solution might be a good approach also for other teams and departments.
Lat: Would you suggest to spread the hackathon culture even more then?
PC: Definitely! Hackathons are a great way to encourage innovation and cross-team collaboration. But it should, of course, remain something special. An engineering highlight once or twice a year, with a change of scenery, so participants can 100% focus on individual ideas, outside their regular work environments. I’m also a big fan of pitching systems for hackathons. As a former engineer, I know how appealing it is to hack on your favorite project, and just “nerd out” for two or three days in a row, but introducing the challenge to also pitch an idea to your colleagues, brings the whole project to a whole new level. You are starting to self-reflect: it’s suddenly not only about the “how”, but also the “why”—Why is my project valuable for me/others/trivago itself? I’m happy to work for a company, which not only embraces but really encourage the culture of hackathons. With over 1500 trivago talents in 2018, it’s sometimes challenging to look outside your own little bubble, and grasp everything that happens left and right of you. But I’m a strong believer in diversity, and I think hackathons, either small or big, are an awesome driver to encourage even more diverse teams and collaboration
Lat: We’re happy to hear that, thanks a lot for the insights and can’t wait for other hackhatons in the future!