WHAT IS HACKING?
A hackathon is usually a competitive event where people work in groups on software or hardware projects, with the goal of creating a functioning product by the time the curtain comes down.
In my experience, beyond solution development, a hackathon also involves pitching the solution to a wider community and canvassing their support through:
1. Eliciting their interest in WHAT good might come out of it
2. Engaging their help by helping them understand WHY you need it
3. Establishing HOW they can help you improve the solution
The pitch is almost as important as the data and analytics, if the solution is to have a real shot at being supported and launched. I’ve participated in about 10 hackathons over the past 8 years, and the experience has sharpened my business problem-solving approach in my everyday work.
Here’s my “3-step recipe” to solving business problems in a lean and fast way.
Step 1: The Prep — Securing sponsorship from senior leadership
Organising a hackathon is expensive. Beyond the prize money or the promise of funding, there are also costs involved for publicity, scouting for relevant participants, running the event, and more. Similarly, launching a data science project is a significant commitment in terms of resources. Thus, sponsorship from senior stakeholders (Directors or, ideally, Vice Presidents or above), in the form of funding, manpower, and time, is a demonstration of their commitment to seeing the project through. I typically seek out senior management support with a proposal guided by this acronym — FETCH:
1. Funding — Secure the money to support the project (go for a small amount first, prove value, then ask for more)
2. Engagement — Set up bi-weekly or, at a minimum, monthly review
3. Team — Team up with subject matter experts from relevant functions to help co-develop the solution
4. Change management — Transform the ways of working within the organisation
5. Handover — Facilitate the transition of ownership to the stakeholder’s organisation when the solution is “business as usual”
Step 2: The Squad — Building the A-team
After securing my stakeholders’ support, the next step is to put together the right team. Like managing a sports team, I have to ensure that each teammate brings along the right set of skills. It is critical to get the mix of technical skills right, from data engineering, data science to application development, and this is very much dependent on the scope and resources available.
· Data engineers are responsible for data acquisition and management
· Data scientists are responsible for performing data mining and modelling
· Application developers are responsible for building a great user experience via visualisation and interface.
One other thing: With all these focus on technical expertise, it is easy to lose sight of the importance of bringing onboard people with the necessary soft skills. For example, recruiting team members who are able to communicate well, work with agility, and learn fast can be even harder than finding those who have the technical skills.
Here’s a good rule of thumb — at least one member of the team should possess excellent communication and business understanding that will help bridge the world of technological solutions and commercial needs. We call these people unicorns and they are often found within the organisation’s functional areas. They love how technology can help solve business problems and they have the capability to bring together business context, understanding, and influence in a way that can vastly accelerate the development cycles and help land the solution with the broader organisation.
Step 3: The Approach — Speed is everything
The world is changing too rapidly for any organisation to take things slow. However, it is still critical to balance the desire for fast iteration with high quality output, and to keep your hackathon mentors/business stakeholders engaged on an on-going basis. It is wise to plan ahead and set aside time for regular check-ins with them. Speaking from experience, it usually pays off in the long run if you show that you are mindful of the stakeholders’ busy daily schedules, like conducting the conversation in a business outcome-focused manner (be prepared, concise, and clear on actions and help needed).
Remember to be realistic on what can or cannot be done in the given time frame too, while exuding a sense of urgency to get started and to get going. Structure your functionality development into a crawl/walk/run manner to continuously show progress towards the business objectives. Starting small on the scope and doing it well to prove or test a concept can lead to a higher chance of success, as opposed to boiling the ocean via overcommitment. To achieve this, reduce the amount of important KPIs or work on the top few brands or customers to show what could be possible. In this way, you can reduce the risk of losing your stakeholders’ interest and patience, when they don’t see progress over a prolonged period.
With the project team, it is recommended to set up regular check-in sessions, such as a daily stand-up meeting, to ensure that everyone is working towards a common goal with maximum efficiency. This communication is especially useful in data science because product development is sequential in nature — it’s not about parallel efforts and deliverables. It is important that the data science team doesn’t become too engrossed in development and technical work, to a point where they neglect the importance of preparing for stakeholder engagement. The presentation of the solution is equally important as solution development. The data science team should try to leave ample time and energy at the end of solution development to think through how they will land the capability with their users. This requires significant discussion and support from their stakeholders — launching and leaving simply won’t work.
How Unilever unlocks these opportunities for data scientists
I am glad that Unilever has been a platform for me to “hack” business challenges on a regular basis and more. There’s plenty of joy and a great sense of fulfilment in seeing the impact of my work for our business teams. For example, in 2018, I got to work on a sales recommendation solution to increase distribution capability for the sales teams. My team and I managed to complete the project as a lean team of four, in less than four months. The project was sponsored by the Vice President of Market Sales and we ran through the 3-step recipe to deliver impact from end to end. Though it started out as a small concept, developed locally on a sandbox, the pilot quickly showed value and user enthusiasm for its capability to identify new growth opportunities at a micro level. This helped garner support to scale the solution to 10,000s of outlets through delivery via a mobile app. In less than 2 business quarters, we were able to go from concept to scale, which resulted in incremental sales growth driven by data.
This example, along with many others, is a result of Unilever being a strong advocate of collaboration with a wide spectrum of data talent and solution providers. We frequently partner with established companies, universities, start-ups and individual professionals to accelerate our data science agenda. I’ve recently been appointed the technical lead for Unilever China’s first hackathon, “AI Change U”. This has enabled me to leverage my external and internal hackathon experience to match what’s possible with what’s needed by my key stakeholders, while creating opportunities for external partners to help us achieve our business ambitions.
If you have the desire to take on important business problems, enjoy working on new data science challenges every now and then, and love solving them in an innovative and impactful way, Unilever might just be the right place for you to embark on your data science career. Check out our available jobs here.
By Kelvin Lim and Eric Chen from Unilever’s Information & Analytics team: