For most of my career, I believed what many finance professionals believe: if a forecast was not good enough, the answer was usually more detail. More drivers. More assumptions. More scenarios. More granularity. More precision. It seemed logical. Business environments were becoming increasingly complex, so forecasting models needed to become more sophisticated to keep up.

Then I watched a forecasting model become so sophisticated that it stopped helping us make decisions. The model was not broken. It was accurate. That was the problem. Several years ago, I was involved in liquidity forecasting during a period of significant volatility in the banking sector. Deposit behaviour became increasingly unpredictable, regulatory liquidity thresholds were under pressure, and management needed answers quickly.

Like many forecasting systems, our model evolved gradually. Every change was reasonable. Additional assumptions improved accuracy. New scenarios captured emerging risks. More detailed segmentation provided deeper insight. Nobody ever proposed making the model unnecessarily complicated. Complexity arrived one sensible improvement at a time.

Eventually, the model grew into a forecasting engine containing thousands of rows, numerous dependencies, and enough assumptions to explain almost every possible outcome. From a technical perspective, it was impressive. From a business perspective, it was becoming a liability. Updating the model could take nearly two days. By the time the latest forecast was available, market conditions had often shifted again. Leadership discussions about liquidity risk sometimes occurred before the model had finished producing answers. Finance teams spent increasing amounts of time maintaining the forecasting process and less time helping leaders evaluate options.

At first, we treated this as an operational problem. Perhaps the model needed better automation. Perhaps we needed faster systems. Perhaps we needed more resources.

But eventually I realised we were asking the wrong question. The problem was not that the model was too slow. The problem was that we had designed it for a different objective. We had optimised for analytical completeness when the business actually needed decision support.

That realisation led me to challenge one of the most deeply embedded assumptions in FP&A: the belief that more complexity naturally produces better decisions. In reality, the relationship is often far less straightforward. Most forecasting models do not become ineffective because they lack information. They become ineffective because they contain too much of it.

Complexity rarely arrives all at once. It accumulates gradually over the years. A forecast misses expectations, so another driver is added. A business leader requests more detail, so additional categories are introduced. A new data source becomes available, so it is incorporated into the model. Every decision appears justified at the time.

The problem only becomes visible later, when nobody remembers why half of the assumptions exist. Many organisations eventually find themselves maintaining forecasting models that resemble historical archives of every concern, request, and business challenge they have ever faced. The result is a model that reflects everything the organisation has learned, but not necessarily everything it needs.

This happens because finance professionals often confuse complexity with rigour. A larger model feels more professional than a smaller one. A forecast supported by dozens of assumptions appears more credible than one supported by five. Precision creates a sense of control. But precision and usefulness are not the same thing. Leadership teams rarely ask whether a model contains enough variables. They ask different questions. What does this mean? What should we do? How quickly can we decide?

Those questions expose a reality that many forecasting processes ignore: forecasts themselves do not create value. Decisions do. This is where FP&A leadership matters. Finance teams should not only build forecasting models but also regularly challenge whether those models still serve the decisions they were originally designed to support. One of the most valuable roles of FP&A is knowing when to simplify, not only when to add more analysis.

A perfectly accurate forecast that arrives too late creates little value. A sophisticated forecast that leaders cannot explain creates little value. A model that overwhelms decision-makers with detail creates little value regardless of how statistically robust it may be. At some point, forecasting stops being an analytical exercise and becomes a practical one. The question is no longer whether the model is correct. The question is whether the model is useful.

This idea became the foundation of what I now call the Minimum Effective Model, or MEM. The Minimum Effective Model is not the simplest model possible. MEM represents the minimum amount of complexity required to support a specific decision with confidence. As an analytical tool, it encourages teams to ask a series of practical questions: What decision is the model supporting? How quickly must that decision be made? Which variables truly drive the outcome? What level of accuracy is sufficient to act? And what complexity exists simply because it has always been there? While the answers vary across organisations, the objective remains constant: to identify the minimum level of complexity required to support a high-quality decision.

That distinction matters because it changes the purpose of forecasting. Once the objective becomes to improve decisions, complexity must justify its existence. Every assumption, scenario, calculation, and layer of detail should answer a simple question: Does this improve the quality of the decision? If it does, it belongs in the model. If it does not, it may simply be consuming time, attention, and organisational energy without creating corresponding value.

This shift sounds subtle, but it changes how forecasting is approached. For FP&A leaders, this represents a different way of thinking. Instead of asking, “How can we make the model more accurate?” the question becomes, “How accurate does the model need to be for us to act?” Instead of rewarding comprehensiveness, organisations begin valuing clarity. Instead of building models around available data, they begin building models around decisions. The result is no less rigorous. It is rigour applied more intentionally.

One practical way to apply this thinking is to test the forecasting complexity by asking the questions FP&A asks during model design and review.

Table 1: A Simple FP&A Test for Forecasting Complexity

Ironically, reducing unnecessary complexity often improves the influence of FP&A. Leadership discussions become more focused. Scenario analysis becomes faster. Decision-makers engage more directly with the forecasting process because they understand the logic behind it. Finance teams spend less time defending assumptions and more time discussing trade-offs.

The conversation shifts from explaining spreadsheets to evaluating choices. That is where FP&A creates the greatest value. This challenge is becoming even more important as organisations adopt increasingly advanced analytical tools and Artificial Intelligence. For the first time, creating larger and more sophisticated models is becoming easier rather than harder.

But easier model creation does not automatically produce better decisions. In fact, it may produce the opposite. The next generation of forecasting challenges is unlikely to come from a shortage of data. Most organisations already have more data than they can effectively use. The challenge will be deciding what deserves attention and what does not. In a world where analysis becomes abundant, judgment becomes scarce.

Organisations that succeed will not necessarily be those with the largest forecasting models or the most sophisticated algorithms. They will be the ones that understand a simple principle: The value of a forecast is not determined by how much information it contains. It is determined by whether it helps someone make a better decision while there is still time to act. Sometimes that requires more analysis. But often it requires something much harder: Knowing when enough is enough.