Introduction

Today, FP&A (Financial Planning and Analysis) teams spend too much time (46%) on low-value activities such as managing multiple spreadsheets and data cleansing, reconciliation, and validation, and do not spend enough time on high-value activities such as scenario planning and change implementation [2]. Hence, progressive FP&A teams are increasingly looking at harnessing digital FP&A platforms for process automation, data, and analytics, including Artificial Intelligence and Machine Learning (AI and ML), to derive insights and make decisions to improve business performance. Basically, AI/ML solutions offer the ability to solve specific problems by using vast volumes of quality data without human intervention. The AI/ML technologies are powered by abundant data captured by digital platforms, along with the availability of secure, cost-effective, and reliable storage and processing offered by cloud computing solutions [2].

Drivers for AI/ML in FP&A

Today, the data economy is becoming increasingly embraced worldwide. Data has enabled firms such as Netflix, Google, Tesla, Lego, Novartis, and Uber to acquire a distinct competitive advantage. Fundamentally, companies that are data-driven demonstrate improved business performance. A report from MIT found that digitally mature firms are 26% more profitable than their peers. McKinsey Global Institute indicates that data-driven organisations are 23 times more likely to acquire customers, six times as likely to retain customers, and 19 times more profitable. According to the Industry Analyst firm Forrester, organisations that use data to derive insights for decision-making are almost three times more likely to achieve double-digit growth [3]. Overall, the drivers for AI/ML in FP&A are increasing revenue, reducing costs, and mitigating business risks. 

FP&A Scenarios for AI/ML

Broadly speaking, AI/ML can support three important business scenarios [4]:

• automating business processes,

• gaining insight through data analysis, 

• engaging with stakeholders.

In process automation, AI/ML can easily and quickly emulate human actions while interacting with digital systems based on predefined business and data rules. This could cover areas such as evaluation of business risk, credit rating calculation, approving invoices, updating customer records in the CRM application, managing Zero-Based Budgeting (ZBB), and more. AI/ML can also help to derive insights and make decisions, including making changes to the systems. AI/ML can cover any areas of Descriptive Analytics, i.e., what happened in the past business performance, Predictive Analytics, i.e., what will happen in future business performance, and Perspective Analytics, i.e., what are the events and factors that will result in a particular outcome (including scenario planning and sensitivity analysis). Last but not least, AI/ML can help in engaging with stakeholders, including customers, vendors, and employees, while managing assets, increasing sales, optimizing product prices, and more. 

Case Study of AI/ML in Digital FP&A

At its core, digital FP&A focuses on process automation, agility, transparency, security, scale, and metric-based insights. With this backdrop, let’s take the most popular FP&A use case - Revenue Forecasting, and explore how AI/ML can help in better forecasting the revenue for a retail chain. Revenue Forecasting in most companies is typically done using the data coming from accounting transactions such as sales orders, deliveries, and invoices. Predicting the revenue using only the first-party accounting data is incomplete if the model doesn’t use other important kinds of data. The first-party data needs to be further augmented with leads (sales funnel) data, forecast data from the channel partners, third-party weather data, inflation data, and so on. When the data from different sources is integrated, that data can be fed into Ensemble revenue forecasting AI/ML model that uses a combination of algorithms such as Linear/Logistics Regression, Random Forest Regression, XGBoost Long Short-Term Memory (LSTM), Autoregressive Integrated Moving Average (ARIMA) Time Series Forecasting, and more. Ensemble modelling is a process where multiple diverse algorithms are used to predict an outcome by aggregating the predictions of each base model into one final prediction.

Achieving Quality Data for AI/ML

However, the accuracy of these AI/ML models depends on the availability of large amounts of quality or trusted data in a single source. In 2025, only 17% of respondents report having either best-in-class or advanced-level data quality. At the same time, 11% of firms still lack even a single source of truth that everyone in the organisation trusts. [1]. So, what can the FP&A teams do to achieve better quality data? Quality data in the AI/ML context is based on three key factors [5]. 

1. Data Dimensions

Firstly, AI/ML solutions need the right data dimensions.  The insights derived from AI/ML are dependent on the response (effect) and explanatory (cause) variables, and these variables are known as features or dimensions. Basically, there has to be the right dimensional data associated with the business process to derive the insights/KPIs.

2. Data Structure

The right data structure holds the key to AI/ML models. Up to 80% of the data captured in business enterprises is unstructured data or TAVI data (text, audio, video, and images) [2]. While this type of unstructured data is easy to capture, it has very little value for AI/ML algorithms as it lacks a predefined data model required for data analysis and processing. Hence, the data for AI/ML needs to be structured for efficient querying.

3. Process Variation

Lastly, most business processes, including accounting, inherently have some degree of process variation, and this variation is reflected in the data that is captured. But if this variation is large, it makes it difficult for the AI/ML algorithms to derive quality and reliable insights. So, the data to be used in AI/ML should have low variability.

At the same time, while quality data is a key enabler for AI/ML success in FP&A teams, quality data alone will not be sufficient. Successful deployment of AI/ML solutions depends on supplementing quality data with the right culture, literacy, technology, and governance.

Conclusion

AI/ML is poised to transform the way FP&A teams conduct business in the coming years. The recent business environment dictates that FP&A needs to shift into the role of strategic advisor in order to transform the way that decisions are made in business. Fundamentally, realising the AI/ML capabilities in business is a journey in the measurement continuum and is not a one-time project. Hence, in today’s VUCA (volatility, uncertainty, complexity, and ambiguity) world, the FP&A teams have to think innovatively, adopt digital FP&A solutions, and contribute more to improve the business performance by increasing the revenue, reducing costs, and mitigating business risks in near-real time. 

References

1. FP&A Trends Survey, "From Ambition to Execution: How Leading FP&A Teams Turn Insights into Impact", 2025

2. Southekal, Prashanth, “Data Quality: Empowering businesses with Analytics and AI”, John Wiley, 2023

3. Southekal, Prashanth, “Analytics Best Practices”, Technics, 2020

4. https://hbr.org/2018/01/artificial-intelligence-for-the-real-world

5. https://www.forbes.com/sites/forbestechcouncil/2021/03/12/three-key-conditions-for-robust-ml-solutions/?sh=2496c4ba2024