By Concentric, Apr 19, 2021
When businesses want to promote collaboration and spark innovation, a hackathon is usually where they turn.
Writing for the American Psychological Association, Shawn Rhoads of the APA Science Student Council explained that these semi-structured events are meant to bring researchers and developers from across a company together to collaborate and solve complex problems. Hackathons usually span multiple days and are extremely flexible. They offer teams who may not collaborate much in the course of everyday work an opportunity to put their heads together to solve a defined issue of common interest.
Recently, Concentric had a hackathon to perfect a difficult algorithm used in our software. Now we’re pulling back the curtain to show you exactly how everything went down and provide some insight into the results.
As you probably imagine, organizations hold hackathons for several reasons. Maybe it’s a health care institution tackling how to bring treatments to rural areas or a national bank trying to find a better way to analyze consumer credit scores. Each industry has its unique challenges it is trying to solve to improve the customer experience or remain competitive.
For Concentric, the reason for our hackathon was to workshop a new algorithm to enrich the capabilities of our software. The goal was to enable users to see how they would need to change their marketing plan to hit their sales goal. For instance, if an organization had a revenue goal of $500 million a year, this algorithm would help them identify the most efficient marketing plan to meet it. Perhaps the team would need to redistribute marketing investments or diversify advertising channels. No matter what, our new algorithm needed to quickly return a good answer to their question.
Of course, by “good,” we mean the recommendation that would enable the user to reach their sales goal efficiently from an investment perspective. Our engineers, developers, and data scientists understand that while there’s no single best solution, working together to solve this complex problem will help us come up with the algorithm that efficiently solves our users’ business objectives.
With a clear understanding of our goal, the hackathon at Concentric commenced. These events are typically associated with a high-pressure, high-stakes atmosphere, with teams stuck in rooms for hours on end trying to solve a problem. However, this is not how Concentric approaches things.
Just as collaboration is the foundation of our platform, it’s also what drove our hackathon. Splitting into multiple teams gave us a construct to avoid quick convergence on a single solution, a result that is more likely if a general brainstorming session is used. It also allows data science and tech team members to work together while bringing their individual strengths to the forefront.
Each cross-functional team pools their talents and diversity of thought to explore different ideas and paths toward a solution.
Our hackathon didn’t lose its competitive edge, however. We balance ambitious and teamwork-oriented forces:
Competitive: Each team is propelled by a drive to create the best possible algorithm for the situation.
Collaborative: The entire process, from creating teams to brainstorming solutions, is iterative and transparent. Each team shares its algorithms, results, findings, and code after each iteration. Teams are allowed and fully encouraged to adopt and build on top of other teams’ approaches.
Concentric’s hackathon was about efficient and effective collaboration and served as a learning experience for every participant. The best solution is one that is cross-pollinated by many teams and iterations, making it ideal for our software and our customers.
Variability is a major component of our software. It’s a major challenge to make an algorithm robust to the randomness in our solution. But our data scientists, engineers, and developers were up for the challenge of finding results from many probabilistic draws.
Our teams don’t shy away from uncertainty and understand that it’s part of creating the most effective solution. Concentric maps randomness to find an algorithm that helps business users understand their best marketing investment across a range of possible outcomes.
Randomness is a given in a hackathon, which means teams need to break out of the idea that success is linear. As developers and data scientists collaborate, they must open their minds to new ideas and adapt their solutions as new information becomes available from internal discussions and other teams. By the end of this robust and flexible process, we should have an algorithm that performs well in all circumstances, which benefits our users the most.
By the time Concentric’s hackathon came to a close, our teams accomplished what we aimed to achieve. We successfully developed an algorithm that shows users how to optimize their marketing spend to achieve their sales goals.
A balance of competition and collaboration is exactly what led us to the best performing algorithm — the winner borrowed heavily from another high-performing competitor. By testing dramatic reallocations of spend based on the current performance of each marketing touchpoint, the new algorithm is able to identify where the return on advertising spend coverages, making adjustments to find the optimal allocation.
As users work to understand their best marketing investment across several business outcomes, the algorithm helps to ensure spending is optimized across all channels.
While we set out what we planned to do, we also were able to grow our own expertise through collaborative experimentation and some lively discussions. Breaking traditional assumptions on how business should be done is what we do at Concentric, and the hackathon only continued to prove the power of thinking outside the box.