Forecasting cash flow with predictive modeling

Understanding business finances in both the short and long-term is crucial for business growth. That’s why a reliable cash flow forecast is so important for businesses in any industry. But for complex businesses with multiple employees, suppliers, vendors and partners, predictive cash flow modeling is especially vital to keep work moving forward.

The Problem: Accurate cash flow modeling

A nationwide mechanical contractor specializing in plumbing and HVAC was looking for a better way to manage projects and cash flow. With multiple variables, including project location, build type (such as condominiums or single-story buildings), type of work being provided (HVAC or plumbing) and estimated project duration, it was difficult for the company to forecast monthly project cash flow.

Accurate cash flow modeling is vital for both general and subcontractors. It impacts key business areas including resource planning and budgeting, the ability to secure loans, progress monitoring, forecasting overall business performance, and stakeholder management.

Discrepancies between actual and forecasted cash flows were requiring extra resources and slowing things down. Project planners were utilizing Excel documents that provided cash flow projections based largely on overly-simplified linear regression models, and they had to update them manually any time cash flow changes occurred. This was even more difficult because payments were usually reliant on hitting certain staged milestones, so numerous data points had to be updated whenever there was a single delay or change. Company leaders understood project managers were utilizing too much time and energy keeping an existing, ad-hoc system full of flaws going. It was time for an upgrade.

The Solution: Predictive cash flow model

Vertex Intelligence, a data science company, began working with the mechanical contractor to develop a predictive model that would improve the ability to forecast project cash flow, saving time and resources.

Typical machine learning approaches to time-series forecasting require thousands or even millions of observations depending on the complexity of the problem. The client only had 100 completed projects as the dataset, which made forecasting cash flow with accuracy even more challenging. To overcome this, Vertex Intelligence leveraged its experience with statistical modeling to restructure the problem itself.

Instead of predicting cash flow as a function of time, Vertex Intelligence developed a parametric statistical model for the cumulative revenue percentage collected over the lifetime of the project. By restructuring the problem utilizing a Hierarchical Bayesian model, Vertex Intelligence could take into account complex relationships between location, build type, trade and project duration as individual inputs.

Results: increased the client’s cash flow forecasting accuracy by 47%.

After analyzing the previous predictive modeling method and all available data points, Vertex Intelligence was able to provide the client with a new forecasting cash flow tool in a matter of days. The new tool saved them time and increased their cash flow forecasting accuracy by 47%.

Although the data science methods utilized for this project were very computationally intensive, the results could be pre-computed and packaged in a simple, familiar Excel spreadsheet. This allowed the user to enter in budget and timeline information and select project parameters with drop down menus they were already accustomed to using. The algorithms and methods for predictive modeling were all abstracted from the user, and cash flow forecasts and charts were automatically updated.

Now, rather than spending hours and days developing cash flow forecasts for a single project, the client builds highly-accurate forecasts tailored to the specific details of the project in mere seconds.

With the help of Vertex Intelligence, this client is able to spend more time on everyday tasks and no longer has to worry about the headache forecasting cash flow inaccuracies previously provided. Because the project was so successful, the contractor is once again tapping Vertex Intelligence to integrate the solution into their CRM to allow for real-time updates to forecasts.