In many companies, the finance department is moving from a transactional to an analytical approach. This is certainly the case for renowned developer and manufacturer of pioneering audio solutions Bose, who has spent the last few years investing heavily in creating a synergy between finance and data science. But how did this come about? Financial Advisor for European Wholesale Wim Theuwissen, Financial & Business Advisor Maarten Werckx, and Senior Data Scientist Vincent Mullers give us the full lowdown.
Bose literally sets the tone in the world of audio equipment and wearables. The company was founded in 1964 by Dr. Amar Bose, an American engineer of Bangladeshi origin who was passionate about conducting research into acoustics and speaker design. Over recent decades, Bose has grown into a pioneering manufacturer of wide-ranging audio solutions for both recreational and professional use. Today, the Bose Corporation has branches all over the world. With technological innovation and sustainability at the heart of this multinational’s DNA, its employees are genuine pioneers in the world of sound.
Rapidly evolving finance department
At Bose – whose Belgian branch is located in Tongeren – the finance department has undergone a total makeover over the past few years, moving from a transactional to an analytical-based approach, setting up a Shared Service Center (SSC) in Portugal. Country controllers in the company work closely with the SSC for their statutory annual financial statements, while its finance navigation team is responsible for all management reports, including forecasting, budgeting, and reporting.
To perform this array of analytical tasks as efficiently and accurately as possible, while at the same time providing maximum added value for the business, the finance navigation team makes use of data science. More specifically, this means using statistical and scientifically substantiated analyses based on large datasets – including external data – to get a clear picture of the business and to assess new opportunities as accurately as possible. Bose has an internal data science team whose job it is to collect data and build predictive models.
From silos to ‘data democratization’
It was about four years ago that Bose embarked on its journey towards having a professionally developed approach to data science. Back then, the business’s data was split between different silos, meaning its availability across the organization was somewhat fragmented. This made it difficult to find out who used what data, while not everyone had access to all possible tools and systems (including reporting systems). This ultimately lead to a difference in perceived realities when analyzing and reporting business results, while also affecting any subsequent (strategic) decision making.
Today, Bose is sparing no expense in centralizing its data and data models. Based on various calculation and reporting tools, management is aiming to develop one central tool across all different silos. The creation of a sales data model is also a work in progress.
The goal is to achieve ‘data democratization’, resulting in every Bose employee having easy access to company data. This should be made possible through centralizing data models, data retrieval by data engineering teams, and creating models via a data analytics team.
Achieving ‘data democratization’, will give every Bose employee easy access to company data.
Data science at Bose: some examples
At Bose, the use of data science has already yielded many useful business insights, with the organization boasting inspiring examples in the areas of finance, sales, marketing, and supply chain.
For example, the use of data science within the finance department gave that extra push towards the introduction of rolling forecasts, or progressive forecasting as a supplement to traditional budgets.
For marketing, a project was launched related to the cost of displays in external stores compared to their revenue. Many data points were used for this project: Who clicks where on which display? What is Bose’s revenue per store? And so on. Based on the results collected, employees were able to start improving the way they configured their displays.
In sales, several one-off analyses were made on, for example, the impact of product color variations on revenue forecasts, the impact of sales via large retail chains on Bose’s own in-store sales, or the effect of promotional pricing on annual or regional results.
The data scientist skill set
Can just about anyone become a data scientist? Or does this role require a specific skill set to make a difference for your business? A data scientist often has a master’s degree in finance and applied mathematics, usually supplemented with a thorough knowledge of programming and AI technologies. Or, as Vincent so aptly puts it: “A data scientist must be better at programming than a statistician and better at statistics than a computer scientist.”
Another essential skill is the ability to properly assess which solutions to deploy for business questions, in such a way that is also accessible for internal end customers. Here, there is an obvious link to having good people skills. A good data scientist must be able to explain complex results in everyday language to coworkers with less technical expertise. This means that empathy, clear communication, and good presentation skills are all vital elements of any talented data scientist’s skill set.
A data scientist must be better at programming than a statistician and better at statistics than a computer scientist.
Last but not least, having the power of persuasion is also an important asset when it comes to this role. It enables you to convince as many coworkers as possible to use data science solutions as effectively as possible. At the same time, an extra push from senior management is always welcome! After all, that is how you build up a good track record of success.
Possible pitfalls of data science
Every business owner and data scientist at Bose agrees that there are several hidden pitfalls when it comes to applying data science. One of them includes insufficient analysis and prioritization of business cases. This element is crucial to ensuring sufficient buy-in from senior management.
Another possible pitfall might include lack of ownership. Any data science project must be promoted enough internally, so that its results also become effective within the organization. In all circumstances, data scientists must avoid spending time, resources, and energy on projects that no one uses.
Another important point of concern is the use of sufficient and reliable data in data science projects. In certain cases, a lot of data is needed in order to measure potential effects with sufficient credibility. However, there may also be times where less complex solutions than data science are required.
Prioritizing business cases is crucial to ensuring sufficient buy-in from senior management.
The future of finance at Bose
Wim, Maarten, and Vincent all believe that the importance of both finance professionals and data scientists within multinationals such as Bose is only set to continue growing over the coming years. The way they see it, the continuing rise of AI and machine learning will create new dynamics, enabling finance teams to spend more time focusing on monitoring and consultancy work. At the same time, the growing use of mathematical models will make for better quality forecasts, much to the increased satisfaction of internal business customers.
