Data Science for the CFO: How data science can help sweat company assets

Every company is different, but most businesses have some common areas that they should consider when transitioning to data-centric ways of working. This is the first in a series of blogs where I’ll present some ways in which the various functions of a business (including Finance, HR and Sales & Marketing) can incorporate data science to improve efficiency and streamline their operations.

Using analytics to improve corporate finance:

CFOs and Finance Directors are sometimes rarely thought of in data science, which tends to concentrate heavily on the operating and customer sides of businesses. However, modern finance leaders are often strong influencers in the boardroom, and their buy-in can be critical to the success of data science within a corporation. To achieve this, it is vital for data science to articulate use cases that speak to the priorities of finance executives. One path can be to articulate ways for data science to help improve a company’s asset base. Two of the largest assets on any company’s balance sheet are inventories and property, plant and equipment (PPE). Often these can be utilised more effectively using data.

The real estate footprint is a major expense for many companies. Though minimising it will be dependent to a large degree on the nature of the company’s operations and strategic decisions about which markets to compete in, a data driven approach can play an important role too. For retailers with large numbers of physical stores, this can be critical. By collecting large amounts of open Geographic Information Systems (GIS) data about building specifics, footfall/traffic, relative locations of local amenities, it is possible to build a very detailed prediction of the exact revenue estimates at individual location. Pret a Manger use data to identify that corner stores with lots of light were undervalued by landlords relative to expected footfall[1]. The exact layout of those stores can also be analysed using data, with experiments undertaken to analyse the how differences in layout can affect revenue, even at a very granular level (e.g. placement of a certain SKU on one side of a display versus the other).

Additionally, physical office space is a heavy financial burden for many companies. The pandemic has only accentuated the need to locate premises efficiently to maximise value. To give just one example, I was once involved in a project for a large public sector organisation, optimising the location of their physical estate, both at local and national levels, to minimise travel times and maximise convenience for tens of thousands of staff and millions of potential service users.

Within the office itself, optimisation algorithms can be used to efficiently plan and schedule the workforce and maximise office space utilisation. This can be done while accounting for factors such as the preferences for each employee in the company on each day of the week, as well as employees who wish to work from multiple locations as well as their home, and the requirements of teams to sit close together and occupy breakout areas. Rather than random hot-desking and the risk of commuting only to find the office out of desks, and teams scattered around, it is possible to take a more structured and efficient approach, and maximum the utilisation of each square foot of space.

Companies with large industrial equipment expenses could take a related approach with plant & equipment as well, optimising the location of capital goods according to factors such as distance to market, the cost and location of existing supply chains, risk of disruption, and so on, to maximise risk-adjusted utilisation of the machinery.

Inventory can be another major area of interest. Inventory is tightly tied to demand. Thus the key to better inventory management (especially of finished goods) can be better demand forecasting. Data science has numerous use cases here. For a retailer, one very obvious one is the use of very granular sales data (e.g. per store, per SKU, per day-level) to improve inventory planning. Deep learning techniques such as recurrent neural networks can be very effective here, as they can find non-linear relationships in the data. A second focus area is reducing the inventory held by optimising the logistics and distribution process. This could involve creating software models of a company’s logistics and distribution process, enabling algorithmic optimisation using data, as well as scenario planning to understand potential bottlenecks, disruption risks, and demand shocks, and the trade-offs that can be made to manage these.

These are just a few of the multitude of ways to improve corporate financial performance using data science. Please get in touch to learn more about ways I can help you to build a data-driven business.

[1] https://www.wired.co.uk/article/pret-barista-subscription-future