When the forecast is always wrong: how CFOs are rebuilding FP&A from the ground up

A mid-sized industrial manufacturer spends six weeks each quarter assembling a rolling forecast. Finance teams consolidate submissions from fourteen business units, reconcile conflicting assumptions, and produce a 47-slide deck that lands on the CFO's desk roughly ten days before the board meeting. By the time the board reviews it, two of the underlying macro assumptions have already shifted. The forecast is technically correct and operationally useless.

This is not an unusual situation. According to research from Gartner (an independent analyst firm), fewer than one in three finance leaders believe their current forecasting process gives them a meaningful advantage in decision-making. The number has barely moved in five years, despite significant investment in planning software and analytics headcount. The problem is not the tools. The problem is the model.

FP&A in 2026: what's actually changing

The structural pressure on FP&A functions has intensified over the past several years, driven by a combination of macro volatility, tighter capital markets, and board-level demand for scenario intelligence rather than point estimates.

Three shifts are worth tracking closely.

First, the planning cycle itself is under pressure. Annual budgets, once the anchor of corporate planning, have lost credibility in organizations operating across volatile input cost environments. Consumer goods companies, energy firms, and any business with significant exposure to currency or commodity markets have largely abandoned the idea that a January budget retains much meaning by August. The response, for leading FP&A functions, has been a move toward shorter rolling horizons, typically 12 to 18 months on a continuous basis, with formal reforecasting triggered by material events rather than calendar dates.

Second, the relationship between FP&A and operational data is changing. Historically, finance teams received data from business units. The direction of travel now runs both ways: FP&A teams at companies like Unilever and Siemens have embedded finance business partners directly into commercial and supply chain functions, so that forecast assumptions are stress-tested in real time against operational signals rather than assembled after the fact. This requires a different skill set. A finance business partner who can read a demand planning model or interrogate a logistics dashboard is substantially more valuable than one who can only consolidate spreadsheets.

Third, AI-assisted forecasting has moved from pilot to production at a meaningful number of large organizations. The honest version of this story is more complicated than vendor marketing suggests. Companies like Anaplan and Workday (both commercial FP&A platform vendors, and their claims should be weighted accordingly) report strong adoption of machine-learning-assisted forecast models. Independent assessments are more mixed. The genuine value tends to show up in specific, bounded use cases: revenue forecasting for businesses with high transaction volume and clean historical data, headcount cost modeling, and working capitalworking capitalWorking capital is the difference between a company's current assets and current liabilities, measuring short-term liquidity and the funds available to run daily operations.View full definition → projections. Broad AI-generated P&L forecasts remain less reliable, particularly in businesses undergoing structural change.

What this means for the CFO

The CFO's role in all of this is less about technology selection and more about organizational design and intellectual honesty.

The most dangerous thing a CFO can do is confuse precision with accuracy. A forecast that runs to two decimal places and is assembled from granular bottom-up inputs can still be systematically wrong if the underlying assumptions are not challenged. The question to ask of any FP&A process is not "how detailed is this?" but "what would have to be true for this forecast to be materially wrong, and how would we know?"

That reframing has practical consequences. It means building scenario infrastructure that is genuinely usable, not a theoretical exercise that finance produces once a year and no one references. CFOs at companies including Philips and Schneider Electric have moved toward three to four actively maintained scenarios, each with its own set of trigger indicators, so that the organization can pivot its operational plan when a specific threshold is crossed rather than waiting for the quarterly reforecast.

It also means being willing to restructure the FP&A team itself. The skills required for a forecasting function oriented around continuous monitoring, scenario management, and business partnership look quite different from those required to run an annual budget process. Many FP&A organizations are carrying significant technical debt in their talent base, not just in their systems.

One underappreciated lever is the governance structure around forecast assumptions. Who owns the macro assumptions that underpin the plan? Who challenges them? At most organizations, these assumptions are inherited from prior cycles or pulled from a single external source without much scrutiny. Building a small, structured process around assumption review, even a two-hour quarterly session with the CFO, head of strategy, and one external perspective, produces disproportionate returns relative to its cost.

Practical steps worth taking now

• MapMapUsing software to automate repetitive marketing tasks and campaigns, enabling personalisation at scale across channels like email, web, and social.View full definition → the latency in your current forecast cycle. Identify specifically where time is lost between data generation and decision-ready output. In most organizations, 60 to 70 percent of cycle time is consumed by data consolidation and reconciliation, not analysis.
• Separate the budget (political, annual, used for performance management) from the forecast (operational, continuous, used for decision-making). Conflating them is the single most common reason FP&A loses credibility with the business.
• Before investing in new planning software, audit the quality of the data feeding into your current tools. A sophisticated planning platform running on inconsistent operational data produces sophisticated-looking noise.
• Define two or three specific forecast metrics you will track against actuals each quarter. Forecast accuracy by business unit, bias direction, and average variance at 90 days out are all measurable. Making accuracy visible creates accountability in a way that process documentation does not.
• Identify one finance business partner role that could be repositioned closer to a high-volatility business function, such as supply chain or commercial pricing, and treat it as a structural experiment before scaling.

The CFO who asks "how do we make our forecast more accurate?" is asking the wrong question. The more useful question is "how do we make our organization faster at responding when the forecast turns out to be wrong?" Building that capability is what separates a competent FP&A function from a strategically valuable one.