How to run a one day Analytics Hackathon that works for your analysts
How is an Analytics Hackathon different from one run for a Data Science, Software Development or Data Engineering team? What makes such an approach work for the different focus & skillsets of analysts?
Three years ago I shared a guest blog post from experienced data leader Ryan den Rooijen about how to run a successful week-long hackathon. It was a popular post with readers. Ryan drew on his experience (then at Dyson) leading mixed teams of analysts, data scientists & data engineers. However, since then some readers have asked me if this would work for a team of analysts in a shorter time period.
So, I was delighted to see this post from Morwenna Causey. Morwenna is an Analytics Manager for Gousto. Podcast listeners may recall that Rob Barham (Director of Data at Gousto) was my guest in episode 29. In that interview he shared about the culture they are building in that business for their different data roles. Through this post, Morwenna brings to life 10 practical considerations to help Analytics leaders run a successful hackathon event with their analysts. I hope it helps you.
Why a Hackathon can help your Analytics team
How often do we think “I could get so much done if I wasn’t in meetings, distracted by slack messages, trying to juggle multiple projects at once.” I know it’s something I think a lot, and also something I hear from my team. So we decided to try something different. Instead of taking an analytics project and having one person working on it over a few weeks, we got a team of data professionals together for one day to come up with as many insights and recommendations as they could!
A hackathon was something we’ve seen in our Software Engineering or Data Science teams, but it was a new concept for driving insights for our Analytics team at Gousto. We set one overarching challenge of unpicking changes in customer behaviour as we come out of Covid. Then split ourselves into 4 teams, each with a different area of focus (and of course cheesy Apprentice style team names were encouraged!)
Here are my chosen ten tips, learned from our first Analytics hackathon…
Top 10 Tips for running an Analytics Hackathon
(1) Preparation is key!
It’s surprising how much preparation was needed for the hackathon. Make sure you allow time for this. This involved defining the brief, gathering context from stakeholders, creating a structure for the day, creating virtual spaces for teams to collaborate and the admin of booking desks, rooms etc. But this meant that in the morning we were able to hit the ground running and jump straight into the analysis.
(2) Make sure your teams are small enough to be focused and collaborate
We started with teams between 4–6 people across different data disciplines and areas of the business to bring different perspectives to the problem space. By keeping these to small groups, each team was able to brainstorm what avenues they wanted to look into and split the work out, while regrouping at regular intervals to update on progress.
(3) Keep the topic broad enough to inspire creativity, but with enough prompts to mean we don’t get lost
We set each team one ‘Starting Question’ but then also added a few prompting questions for them to think about to get the creative juices flowing. We deliberately didn’t want to be too prescriptive with the areas people look into. We’re glad we weren’t as we had some really innovative ideas come out on the day. (I for one didn’t know you could easily access Heathrow passenger numbers…)
However, we also took away the learning that sometimes being too broad can make it difficult to focus on outcomes. Think about articulating to the team what you want as an outcome (eg. what are the specific problem areas you want to solve?) This was something that worked well in a hackathon our Software Engineering team ran previously.
(4) Check in regularly
We had two people whose role was to facilitate the hackathon and make sure everything ran smoothly. This involved checking in on the teams at regular intervals to make sure everyone knew what they were doing and looped in with other teams who might have insights. We also held quick stand ups for every team to give a 2–3min update at the end of the morning and mid afternoon so that everyone heard the highlights.
(5) Keep the energy high!
Energy was mainly fuelled by copious amounts of sugar (and introducing a lot of us to the Polish holiday of Fat Thursday!) We also ran a kick off session in the morning to drive excitement. We had regular check ins to celebrate what we’re learning throughout to help keep energy and motivation levels high right into the afternoon.
(6) Be ready to pivot when you don’t find what you expect
We expected to find some clear trends in each of the topics we set. But this was not the case and one topic to understand how behaviour had differed by various segments of customers actually ended up showing no clear differences. This was a reminder that no differences is a beneficial insight in itself. It can stop you from exploring hypotheses that would have driven differences by customer types and focus on wider macro trends. But it also meant the team had to quickly pivot and try looking into other things when they didn’t find what they initially expected. Allowing teams to be flexible and go in whatever direction the data takes them is key.
(7) More structure is needed when running hybrid hackathons
We had a mixture of people working in our offices and at home and so the ways we collaborated needed to work in a hybrid environment. This involved a Miro board with areas for teams to brainstorm, collect insights and build their story. We also used a slack channel with regular updates, interesting data sources and reminders to keep thinking about that overarching question.
(8) Involve your stakeholders
We tried to gather the information we thought we would need ahead of the hackathon. But it soon became apparent that, as with any Analytics project, the context from our stakeholders was critical. This meant a lot of slack messages asking questions throughout the day as we tried to understand the trends we were seeing! In the next hackathon we’re keen to get more of our stakeholders involved. Both in the context building upfront and also being available for questions through the day.
(9) Plan how you’ll share this back to the business
After the hackathon, our two facilitators took responsibility for pulling all of these insights together into a deck. That is a story that we could share back to the business. Don’t underestimate the time and importance of this step. As well as identifying who are the stakeholders you need on board to drive action from these insights. For us that meant splitting the insights into groups and getting together with stakeholders responsible for those areas to discuss. Then coming up with the next steps to share back with the wider business along with our insights.
(10) Have fun!
This was a great opportunity for us to come together as a team. Plus to work with other people in the data team we don’t usually work closely with. Make sure to keep it light hearted. Enjoy that time together and the break from the day to day. We added a competitive element of who could come up with the best insights to also add to this with prizes at the end of the day.
I hope this post has inspired you to think of running Analytics Hackathons in your teams, I’d love to hear your thoughts and tips in the comments below.
What is your experience of Hackathons for analysts?
Many thanks to Morwenna for sharing her experience of running this event at Gousto. So many of her tips make good sense to me. They ring true with what I’ve also heard works in other businesses & sectors.
What about you? Are you using hackathons for your analytics team? Have you run a successful one-day hackathon? If so, I’d love to also hear your experience and tips for success. It feels like we are still all learning what works best. What is clear is this approach has potential beyond data science teams.
Enjoy your hackathons & I look forward to hearing from you.