No longer historical record-keepers, today’s financial professionals are forward-looking strategic partners to the business. This shift away from a purely retrospective perspective to proactive value creation has brought about demand for a new set of tools. Among these is predictive analytics, which is instrumental for making the leap from answering “what happened?” to “what is likely to happen?”

Developments in artificial intelligence (AI) have advanced what is possible with predictive analytics, and interest among finance leadership in adopting these tools is indicative of their potential to transform the office of the CFO. According to a report from PYMNTS Intelligence and Coupa, 82% of enterprise CFOs are actively using generative AI (GenAI), with over half demonstrating a willingness to invest in AI capabilities for predictive analytics.

This article explores what predictive analytics is and its importance for the finance organization, providing concrete examples and strategies for accounts payable (AP), accounts receivable (AR), treasury and financial planning and analysis (FP&A) functions.

Quick Navigation

• The Analytics Continuum
• Predictive vs. Prescriptive Analytics
• Use Cases for Predictive Analytics in the Finance Organization
• A Strategic Framework for Implementing Predictive Analytics in Finance
• Upskilling for Predictive Analytics in Finance

What Is Predictive Analytics? Meaning, Definition & Importance in Finance

Predictive analytics is a subset of business analytics that combines statistical analysis and machine learning (ML) to analyze datasets and make predictions about a dependent variable. The foresight provided by predictive analytics enables financial professionals to forecast more accurately and manage risk proactively, driving greater strategic value for their organization.

Predictive analytics can be applied to continuous variables, such as the weight of raw materials ruined during production, or discrete variables, such as the number of unsatisfied customers during a quarter.

It’s especially useful in automated decision-making systems that provide event-driven alerts related to unusual events. An example of this is the automated monitoring of credit cards for fraudulent transactions.

The Analytics Continuum

To fully understand predictive analytics, it’s important to see where it fits within the full spectrum of data analytics. The Gartner Analytic Ascendancy Model is a useful framework for understanding the journey from basic reporting to strategic foresight.

1. Descriptive Analytics: “What happened?” Essential, though purely retrospective, this is the foundation of all financial reporting. It includes financial statements, key performance indicator dashboards and sales reports that summarize past performance.
2. Diagnostic Analytics: “Why did it happen?” This step drills down into the data to understand the root causes behind the numbers. An example of diagnostic analytics is using variance analysis to understand why a company spent more on raw materials for a product than was originally budgeted.
3. Predictive Analytics: “What is likely to happen?” This is where the analysis shifts from reactive to proactive. This step uses historical data, statistical algorithms and machine learning to forecast future outcomes.
4. Prescriptive Analytics: “What should we do?” As noted in the AFP Treasury in Practice Guide: Identifying Treasury Value: Automation, Machine Learning, AI, underwritten by Kyriba, this is the most advanced level of analytics. It uses the forecast from predictive analytics to recommend specific optimal actions that will achieve a desired outcome.

For the office of the CFO, moving beyond simply descriptive and diagnostic analytics is a strategic imperative, as it is the difference between being a cost center focused on compliance and a value creation center that shapes company outcomes.

Predictive vs. Prescriptive Analytics

Financial professionals need to understand the differences between predictive and prescriptive analytics to use them most effectively in their work.

Predictive analytics provides foresight. Its outputs include forecasts, probabilities and scores. While it provides a valuable early indicator, the decision on how to use this information to act rests with the financial professional.

Prescriptive analytics provides guidance. Using the forecast from predictive models, it recommends a specific optimal action to the financial professional.

A head-to-head comparison:

Building a robust predictive analytics capability is the first step toward effectively leveraging prescriptive analytics.

Use Cases for Predictive Analytics in the Finance Organization

Leading finance organizations are using predictive analytics to enhance forecasting and scenario planning, optimize cash flow and mitigate risk. Below are examples of predictive analytics use cases for AP, AR, treasury and FP&A functions.

Accounts Payable: Optimizing Outflows and Detecting Payments Fraud

In the AP function, predictive analytics can be used to forecast future invoice volumes by analyzing historical patterns of invoice submissions. This enables AP managers to allocate staff resources optimally, preventing bottlenecks.

Predictive analytics can also help AP teams determine the optimal time to pay suppliers by analyzing vendor invoices, payment terms and historical payment data. A predictive analytics model can determine the statistical likelihood that a supplier will accept an early payment in exchange for a discount at a specific rate.

In addition, predictive analytics can analyze patterns in past transactions to establish a baseline of normal activity and flag anomalies, such as unusual payment amounts or payments to unapproved vendors, that may indicate fraud.

Accounts Receivable: Predicting Payments and Managing Credit Risk

Predictive analytics can help AR teams analyze each customer’s payment history and invoice details to predict when each invoice will be paid. For example, as described in the AFP Executive Guide: What Does the Future Look Like for Treasury?, underwritten by Wells Fargo, Konica Minolta U.S.A. used AI to predict customers’ payment habits, which improved the company’s view of customer behavior, allowing the collections team to focus on chasing large and complicated delinquent accounts.

Knowing which accounts are at risk of becoming delinquent also enables companies to take targeted, preemptive actions to encourage on-time payment, thereby reducing days sales outstanding.

In addition, predictive models can be used to factor in payment history and customer data for a dynamic assessment of customer creditworthiness. This helps companies set appropriate credit limits for their customers.

FP&A: Leveraging Smarter Forecasting and Scenario Planning

Predictive analytics enables FP&A teams to produce more accurate and nuanced forecasts by analyzing large quantities of data and incorporating multiple internal and external variables.

For example, as described in AFP’s case study on IBM’s enterprise transformation, IBM developed a predictive tool that uses machine learning and AI to accelerate and simplify the financial forecasting process. The company forecasts 70,000 different data points a month using a collection of data science models. Its threshold for forecast accuracy is 95%, and in many areas, it realizes 98% or greater.

Additionally, predictive analytics can be used to run sophisticated “what-if” scenarios by quickly simulating the financial impact of hundreds of potential business scenarios. This provides FP&A teams with a clearer understanding of risks and opportunities.

Treasury: Improving Cash Flow Forecasting and Risk Management

Predictive analytics can help treasury teams forecast cash positions with greater accuracy by analyzing real-time and historical data, as well as market trends. With a more comprehensive picture of inflows and outflows, treasury teams can make informed decisions about deploying cash without taking on liquidity risk.

Predictive analytics can also help treasury teams anticipate fluctuations in foreign exchange (FX) rates. By using historical data and testing it against current data within a tightly controlled perspective, treasury teams can develop a predictive model that helps them refine hedging strategies, minimizing potential negative impacts on the business.

For example, as described in AFP’s case study on ASML’s in-house AI solution, ASML created a fully automated AI model for its purchase FX hedging program, which relies on forecasts of expected U.S. dollar-denominated material intake. The team gathered five years of historical actuals, trained algorithms on the first three years of actuals and tested their accuracy on the last two years of actuals. After a series of refinements, the model was able to recognize patterns and trends in historical actuals and use them to predict future intake, improving forecast accuracy and ultimately reducing USD exposures by $25-50 million monthly.

A Strategic Framework for Implementing Predictive Analytics in Finance

Adopting predictive analytics requires more than just the technology. Below are five best practices for a successful predictive analytics project:

1. Begin with a clear, specific business objective. Analytics projects should align with business goals. Create a SMART goal (specific, measurable, achievable, relevant, time-bound) to ensure the project delivers quantifiable value. As the AFP FP&A Guide to AI-Powered Finance advises, teams should start with a manageable scope and focus on smaller wins before expanding.
2. Clean and consolidate data. The principle of “garbage in, garbage out” applies here, especially because predictive analytics projects have a high threshold for clean data. Without high-quality data, the project will not succeed.
3. Refine the model through testing. Run the predictive model in parallel with the old manual process to compare results and fine-tune the predictive model. A rigorous testing process also builds organizational confidence in the predictive model’s outputs. Ask for feedback from business partners.
4. Maintain human oversight. Keeping humans in the loop is critical for accountability and handling ambiguity. Teams should be empowered to question and override an output from a predictive model when it contradicts business knowledge or domain expertise.
5. Review and test the model after implementation to ensure it does not stray from its benchmarks.

Upskilling for Predictive Analytics in Finance

As predictive analytics enhances finance processes, it also transforms the skill set required of financial professionals. While foundational financial acumen remains necessary, it must be combined with technical know-how, strong data literacy and effective communication skills.

Analytical Skills: The “What and Why”

Financial professionals must have the ability to translate data into meaningful business insights. Skills in this category include:

• Financial and business acumen: Accurately interpreting the outputs of a predictive model requires an understanding of financial principles, business models and key performance indicators. Knowing which questions to ask requires business acumen.
• Statistics fundamentals: Concepts such as probability, variance and regression analysis are necessary to understand the accuracy of a model, identify potential flaws and limitations, and determine how to act on the outputs.
• Critical thinking skills: This includes questioning assumptions, interrogating the outcomes, identifying the root causes of trends identified by the model and coming up with next steps to address the root causes.

Technical Skills: The “How”

While most financial professionals likely won’t be building predictive models themselves, it can be helpful to have a functional understanding of the related tools and technologies, including:

• Data wrangling skills, for example, proficiency in Structured Query Language (SQL), to gather data from multiple sources and clean and organize it.
• Basic knowledge of programming languages, such as Python or R, is optional but allows for more complex modeling. For those without coding skills, an AI-powered process, such as the one described in the AFP article “How Financial Professionals Can Apply GenAI to Data Workflows,” can be used to generate code for data processing needs.
• Proficiency in business intelligence (BI) and data visualization tools is necessary for exploring data and presenting findings from the model.

Soft Skills: The “So What”

Even the most precise data will fail to influence change without soft skills, which include:

• Communication and data visualization skills to turn complex analytical findings into a clear business narrative for non-technical stakeholders that appeals to their concerns. People often view machine-generated analytics as a mysterious black box; good communication skills can help stakeholders understand the operations that produced the analysis.
• Collaboration and business partnering skills to work with teams across the company. Predictive analytics projects, in particular, will likely require collaboration with IT, data scientists and data engineers.

A Culture of Continuous Learning

The skills, tools and methods associated with predictive analytics are constantly evolving, particularly as AI continues to advance. As such, financial professionals must maintain a mindset of continuous learning to stay current. Opportunities for professional development include:

AFP Digital Badge: Analytics for Finance: Lead with confidence with predictive analytics.

AFP FP&A Series: No-Code AI for Finance: A complimentary virtual event taking place August 27, 11:00 AM - 3:00 PM ET, designed to help FP&A professionals harness the power of generative AI, machine learning and intelligent agents.

AFP Treasury in Practice Guide: AI Skills for Treasury: Coming later this year, a guide to the essential AI skills that treasury professionals must develop to stay competitive and drive efficiency. Sign up to be notified when the guide is released.