Powering-up your cash planning process with Machine Learning

Managing cashflow is a critical part of any project-based business. In this customer story we examine how Hunt Electric leveraged machine learning and analytics to remove risk, improve forecast accuracy, and optimize their cash planning process.

“We’ve reinvented our cash planning process with Machine Learning. It has dramatically improved the accuracy of our outlook, which removes risk, and allows us to take on new projects with confidence.” – Ron Somerville, CFO

60 years of sustained growth

Founded in 1965, Hunt Electric is an electrical design, build, and maintenance company based in Bloomington, Minnesota. With 1,400 electricians and 200 office staff, it undertakes large-scale projects across the USA, including major healthcare and educational facilities, complex data centers, solar energy installations, and a host of other contracts. Along with Hunt’s expanding portfolio of projects comes mounting financial management challenges. Like any project based business, robust cash management is critical not just to survival, but mandatory if you wish to grow and take on the dramatic inflows and outflows that come with larger projects.

Big projects bring an appetite for cash

Taking on bigger clients brings with it different vendor dynamics – big clients demand more favorable payment terms – and without robust cash planning you quickly become a bank. Concentration risk in a small number of key projects can also create unacceptable risk across the broader portfolio. Late payments from big customers have an outsized impact and force you to borrow, or delay payment to vendors.

Predicting when customers will pay

To mitigate these risks, it’s critical to understand when customers will pay. Our machine learning model predicts the ‘number of days late’ for each invoice based on the size of the bill, the customer’s track record of paying, the type of job, along with dozens of other input variables. The output is an actionable list of the most at-risk invoices. Not only is this useful for the AR team to proactively check-in with customers, but the projected days late flows directly into the automated cash forecast.

Capitalizing on early payment discounts

Timing payments to suppliers to capitalize on early payment discounts, without jeopardizing cash flow and incurring line of credit costs, is a fine balancing act, but made a lot easier when you have confidence in your cash flow forecast. This was particularly challenging to solve, as it involved calculating potential savings from early payment discounts accurately, while avoiding early payments that could force borrowing during high negative cash flow periods.

Getting payroll predictions right with ML

Comprehensive payroll inclusion is also a crucial input into an accurate cash forecast. The construction industry sees not only cyclical and seasonal fluctuations in the pace of projects and demand for labor, but also the complexities of unionized labor forces, over-time rates, and relatively high workforce attrition. To build a clear picture of anticipated payroll costs we integrated office payroll, field overhead, anticipated payroll growth, and seasonal fluctuations into a machine learning model to predict future labor demands on both active and quoted jobs. Tax obligations and payroll expenses spanning multiple states and jurisdictions was also addressed.

Scenario modeling

Enabling the finance team to model hypothetical scenarios is a natural evolution once the analytics and predictive components are in place. Are we better to pay late and incur a penalty, or pay early with more favorable terms, but leverage credit and incur a cost? Modelling these types of outcomes on the fly can help fine tune your payables decision making.

Looking forward

The construction tide rises and falls with interest rates, and the latest cycle in the post-covid era has been no different. Smaller players, with less access to credit, or those exposed by concentration risk, can quickly come under stress. The implications of those stressed companies can quickly flow up and down stream, making liquidity it even more important and building the case for more robust cash planning processes.

Solution Overview

Client Name: Hunt Electric Corporation

Industry: Construction

Geography: Based in Minnesota – Operating USA

Function: Finance

Business Value Drivers: Reimagined Processes and Predictive Insights

Challenges:

• Forecasting Receivables

• Timing Supplier Payments

• Optimizing Borrowing Costs

• Long Range Cash Planning

Solutions:

• Predicted invoice payment dates

• Late payment risk score

• Fully automated cash forecast

• Predicted labor hours ML model

Impacts:

• Better forecast accuracy and control means reduced risk

• Lowered LOC borrowing costs

• Increased take-up of early payment discounts

• Pro-active account management has reduced aged receivables

• Improved labor planning for quoted and active jobs

Resources:

Download Machine Learning Powered Cashflow Planning

Learn more about our Data Science Service