Redesigning Pearson’s Cash Flow Forecasting Using Artificial Intelligence

Published: December 01, 2018

Redesigning Pearson’s Cash Flow Forecasting Using Artificial Intelligence

To improve Pearson’s cash flow forecasting, the London based learning company is rolling out a new, artificial intelligence based solution. It interfaces with the company’s key finance and ordering systems to provide a detailed cash flow forecast in a single dashboard. Treasury is now leveraging the dashboard as the backbone for its iterative forecasting process, with an eye on constant improvement.

A variety of organisational factors make cash flow forecasting challenging in any large multinational organisation such as Pearson. 

Our process previously involved local teams preparing forecasts to feedback into group treasury on a monthly basis, which they obtained from a variety of sources. The local business partners involved already had a wide variety of other demands on their time and we felt that the process was more time consuming for them than it needed to be.

The implementation of an enterprise-wide ERP system provided a great opportunity to look again at how we could re-imagine the process – simplify and make it more efficient, to allow for more time to be spent on insightful analysis.

Defining the parameters

We began by producing a business case for a change. We set out the business benefit that we hoped to achieve and estimated the time and cost required.  

The starting point was to question the timeframe we should forecast using a bottom-up approach. We decided that a two-month rolling forecast would be sufficient to allow us to make informed decisions about the time periods we could invest cash or would need to borrow for. We aimed for a very high level of accuracy for the first month of the cash flow, since most customers and suppliers have payment terms of 30 days or more, with a little more tolerance for the second month where the underlying sales or purchases may also need to be forecast.

Once we had determined the forecasting timeframe, we carefully examined the drivers behind redesigning the cash flow forecast, as a way of determining the value it could add to the business.

The key question was how important the project was to our ability to maximise the value of treasury to the organisation. The group has an operating cost saving target of £300m to be achieved between 2017 and the end of 2019 and we were concerned that, although the return we expected to achieve on the project was good in percentage terms, it was too small to make a meaningful contribution to this target and may divert resources from other projects which would make a larger contribution to the target.

We felt there needed to be greater motivation than just the direct financial return for undertaking the project and we concluded that the project would provide a facilitator for almost all of our treasury processes including liquidity, foreign exchange, interest rate and counterparty risk management. 

In order to avoid diverting scarce resource from other projects, we considered whether the process could be reformed by using staff dedicated solely to cash forecasting, but concluded that this would be labour-intensive, time-consuming to scale, and expensive. We therefore started to look for technology solutions using partners that could deliver similar benefits, with minimal upheaval and a relatively light resource requirement. 

Fig. 1 - High Level Overview of Solution

Fig 1  High Level Overview of Solution

Finding the right fit

When evaluating solution providers, we were clear with our expectations. We wanted a solution that would:

    To meet our requirements it was important that the teams understood what we were trying to achieve, why and how. We have worked with leading providers in this space to develop a model that meets our criteria. Our solution provides a dynamic dashboard that brings together all the required information in a single window (see Figure 1) and uses a variety of data cleansing and analysis techniques to try to ensure the accuracy of our forecasts.

    Reaping the benefits

    We are now well progressed in developing our forecasts for seven entities, representing a meaningful proportion of our cash flow. Already, we are seeing significant benefits, including increased ability to analyse cash flows, greater insight into business trends and customer behaviours, and increased forecast accuracy. As a result, we have been able to reduce our cash float.

    We also hope that the solution will help us to assess the value of new business trials taking place across the business and the behavioural impact we could expect if they are fully implemented across the group. For example, our Australian business is trialling a new online payment portal for digital sales, which enables them to quickly roll out a wide variety of payment solutions. We now have the tools to help track any changes in cash flows we might see as consumers adopt new payment options.

    Looking ahead, we are keen to ensure we are using the dashboard to its full potential. This not only means adding more entities, but also teaching the AI to achieve our desired goals. Ultimately, you get the most out of the solution by combining the team’s expertise with the processing power and learning capacity of the AI. So, it is important to dedicate resources – from treasury and from the vendor’s side – to work with the AI and establish clear rules it can stick to. That’s where the main value of this solution is found: in the intersection between human and artificial intelligence.   

    James Kelly
    SVP, Group Treasurer, Pearson plc

    James is Treasurer of Pearson plc, a FTSE 100 listed global education company. Since joining Pearson, James has led a number of projects including disposals and subsequent debt liability management exercises, foreign exchange management optimisation and automation of activities. He has previously worked as Treasurer of Associated British Ports, Head of Treasury for Rentokil Initial plc and in treasury roles at Sky and Kingfisher plc.

    Sign up for free to read the full article

    Article Last Updated: May 03, 2024

    Related Content