The age of AI is here, and it’s fundamentally reshaping how we approach financial planning & analysis (FP&A). For years we’ve focused on the data that shows up on traditional financial statements – revenue, expenses, cash flows – when pulling together annual budgets and making forecast projections. Often we have built models around the concept of drivers like “units”, “bookings” and “headcount” to help automate these processes, and explore different scenarios such as “how does a 30% increase in raw material costs impact profitability?” and “how would an investment in increasing employee retention reduce training and opportunity costs?”
Spreadsheet-based Planning Won’t Work for AI
As spreadsheets have been supplemented by powerful FP&A software platforms (such as IBM Planning Analytics with Watson) these models have grown in breadth and scale, enabling organizations to improve the quality of their projections, while reducing the turnaround time for an updated budget or forecast from weeks to days (or even hours!). These qualitative and quantitative improvements are due to FP&A software platforms having the following advantages over spreadsheets:
- The ability to store large amounts of detailed data securely, while still being easily accessible for business users via spreadsheets and web browsers. This is the feature that provides “one version of the truth” and saves time because it only needs to be validated once.
- The ability to automate data loads from ERPs and other source systems eliminates time spent on manual data manipulation in spreadsheets including VLOOKUPs and rolling up SUMs/SUMIFs, and opens the door to more detailed analysis.
- Features that support embedding business logic formulas within the database removes the need to “check your spreadsheet formulas” in case a reference is wrong, or it was “incorrectly copied”.
- User/group-level security and browser-based user interfaces make it easier to engage a wider audience in planning processes. Instead of FP&A being an intermediary (and a bottleneck) departmental managers can submit their headcount and expense requests directly AND see the immediate impact on their P&Ls! Meanwhile FP&A and leadership can see consolidated submissions immediately, shortening review and revision cycles significantly. An additional benefit is that it increases engagement across the organization by empowering people to participate directly in the process.
- Last but not least the availability of all actual and plan financial data – and business logic – at greater levels of detail in a single location is what opens the door to leveraging AI. AI feeds on data, and your secret sauce is YOUR data, the more of it, the better.
Better Data means Better AI
Moving beyond spreadsheets for FP&A is the first, necessary, step. And once that is done, the foundations are in place for incorporating other kinds of data that are relevant and material for FP&A as well as integrated business planning in general, for example:
| Demand data from Sales and Operations Planning (S&OP) | S&OP is often done completely separately from financial planning, which leads to time consuming – and potentially contentious – reconciliation exercises towards the end of planning cycles. Incorporating demand data early in the cycle leads to better planning overall, while reducing the number of meetings needed to “get on the same page”. This is also a key component of any xP&A (eXtended Planning & Analysis) roadmap. |
| Outputs from Climate Risk models | As the frequency of extreme weather events (hurricanes, floods, drought etc) increases exponentially as a result of climate change, climate risk models can be used to quantify the financial impact of such events on the assets of a business by geography, under different climate scenarios. In the 1980s when the number of weather disasters that caused $1 Billion in economic damages averaged 3 per year, it was reasonable to ignore it. But when you consider that the number of $1 Billion+ weather events has grown exponentially over the last 20 years, to 27 in 2024, the chances of your operations being impacted if you have facilities in (say) hurricane and flood prone areas, has gone up significantly. At minimum the possibility should be factored into your edge-case scenarios, in addition to anticipating a rise in the cost of insurance premiums. |
| Data from ESG/Sustainability reports | Many companies already collect decision-useful and material non-financial data in an ESG reporting system like IBM Envizi. These data will vary depending on the industry and business context, but typically include green-house gas emissions, as well as other relevant environmental (water usage, waste generation, etc) and social (employee turnover, demographic etc) metrics. |
These three examples show how non-financial data (demand, climate risk, sustainability) can add value to your financial planning processes by saving time, increasing accuracy, and providing greater insights.
Feed your AI Models!
The second major benefit to collecting these kinds of data is to feed your AI models. Anyone who has played with an AI large language model (LLM) such as ChatGPT has likely been impressed with the way it makes connections across wildly different concepts. For example, I asked an LLM this question: “what happened to General Electric (GE), and how did its competitors do?” and received an in-depth analysis comparing GE with the german powerhouse Siemens that touched on wide-ranging topics including the 2008 mortgage crisis, and GE’s disastrous foray into financial engineering while Siemens doubled down on digital innovation and sustainability priorities.
As this example shows, LLMs can quickly find correlations between data sets, and put them to good use, but only if the data is available to the AI in the first place. If this sounds like an exercise that is too open-ended to be useful we recommend conducting a materiality assessment (check out this blog post to find out more), or by looking at one that your organization may already have done. While hurricanes may not be a concern for your business, changing demographics, water scarcity, supply chain disruption, or something else may be. Identifying metrics for material matters that you can collect, track, and incorporate into your analysis – and AI models – may make all the difference in preparing your business for long term sustained success.
How to Proceed
If spreadsheets are still your primary FP&A tool, but you do not want to miss out on the benefits of AI, STOP! Jumping into AI without a clear picture of how it will benefit you will just be a waste of time and money (let others do the time and money wasting!)
- Work on a roadmap to get your FP&A data and models into an FP&A software platform with a database such as IBM Planning Analytics with Watson. If this is a new experience, we recommend doing this in bite-sized pieces, for example starting with expense planning, then headcount planning, then revenue planning and cash forecasting. Change is hard for an organization and a team, and each step is an opportunity to absorb the improvements that come from automation, and to learn from them.
- Once you have governed, trustworthy data in your platform, you can start experimenting with AI. In fact, IBM Planning Analytics with Watson comes with several AI features built-in and others available as add-ons, including predictive forecasting, natural language querying, and Agentic AI.
- With these solid technical foundations in place, you can start collecting non-financial metrics that will add color and depth to your AI models, which will perform better and better over time as a result.
Contact us here or on LinkedIn to find out how we can help you design an FP&A to AI roadmap customized specifically for you to be successful, or to learn more about our design and delivery services for IBM Planning Analytics with Watson, IBM Envizi Reporting, and IBM Envizi Climate Risk Insights.