How Unilever rebuilt its forecasting on drivers, not budgets

By the mid-2010s, Unilever's FP&A teams were producing detailed annual budgets that consumed months of effort and became unreliable within weeks of publication. The company operates across more than 190 countries, selling roughly 400 product brands into markets with wildly different inflation rates, currency dynamics, and commodity exposures. A single annual number for, say, palm oil costs or pricing in Nigeria was a fiction from the moment it was printed. Finance leadership recognised the core problem: the budget was built around the organisational calendar, not around the variables that actually moved revenue and margin.

The deeper issue was structural. Traditional budgets ask managers to commit to a number. Driver-based models ask them to commit to a theory of how the business works. That shift sounds philosophical, but it has direct operational consequences. If you know that Unilever's gross margingross marginGross margin is the share of revenue left after subtracting the direct cost of producing goods or services, expressed as a percentage of revenue.View full definition → in a given category is primarily a function of commodity input costs, pack size mix, and retail price realisation, you can update the forecast the moment any of those inputs move. You do not need to wait for the next planning cycle.

What Unilever actually did

Unilever's finance transformation, developed progressively through the late 2010s and accelerating into the early 2020s, centred on identifying a manageable set of value drivers for each business unit and hardwiring those drivers into the planning model. The company distinguished between external drivers (commodity indices, consumer price inflation, currency rates) and internal drivers (volume growth assumptions, promotional spend ratios, mix between premium and mass-market SKUs). Each business unit was required to mapmapUsing software to automate repetitive marketing tasks and campaigns, enabling personalisation at scale across channels like email, web, and social.View full definition → which two or three external drivers explained the majority of their cost or revenue variance, then build forecasting logic around those specific variables.

The rolling forecast mechanism replaced the annual budget as the primary planning instrument. Rather than locking in a full-year view in November and defending it through December of the following year, finance teams updated a 12-to-18 month forward view on a quarterly basis. The key discipline was that each quarterly update had to restate the driver assumptions explicitly, not just revise the numbers. If the palm oil price assumption changed, the model showed the cascade through gross profit, promotional budget capacity, and operating income automatically.

The technology and governance side

Unilever invested in Anaplan as its planning platform (Anaplan is a commercial vendor of connected planning software; their own case study materials describe the Unilever implementation, and those figures should be read with that commercial context in mind). The platform allowed multiple business units to work in a connected model rather than in disconnected Excel files, which had been the previous state. More important than the technology choice was the governance change: the quarterly business review moved from a backward-looking variance discussion to a forward-looking driver review. Finance presented the updated driver assumptions first, the forecast second. That sequencing change forced business partners to engage with causality rather than just accepting or disputing a number.

Training investment was substantial. FP&A analysts who had spent careers building bottom-up budget templates had to learn to reason in driver logic, distinguishing between changes that were volume-driven, mix-driven, and price-driven. This is harder than it sounds when you are managing 400 brands across dozens of categories.

The results

Unilever's public financial communications in the early 2020s showed improved planning agility, particularly during the commodity inflation shock of 2021 and 2022. The company was able to implement pricing actions with speed that several analysts noted as faster than peer response, which is at least partly attributable to having live driver models that quantified the margin impact of delayed pricing decisions in real time rather than waiting for month-end reporting. During 2022, Unilever took approximately 11 percent price increases across its portfolio while managing volume decline of around 2 percent, a trade-off the finance team modelled explicitly against driver-linked scenarios.

Specific internal productivity metrics for the FP&A function are not publicly disclosed. The Anaplan case materials cite planning cycle time reductions, but given the vendor source, those figures require independent corroboration. What is visible externally is the consistency of Unilever's investor guidance: the company shifted to communicating in terms of volume growth, price realisation, and underlying operating margin rather than point-in-time budget targets, which reflects the internal planning language of a driver-based model.

What transfers, and where your context differs

The Unilever model is not a plug-and-play template, but several mechanics do transfer across company sizes and industries.

Start with driver identification before touching any technology. The analytical work of mapping which two or three variables explain 70 to 80 percent of your revenue or cost variance in each business unit is the actual intellectual product of this exercise. Many finance teams skip this and buy planning software hoping it will perform the thinking for them. It will not.

The quarterly rolling update rhythm works for most organisations, but the forcing function matters: the update must change something. If your quarterly forecast almost always looks like the annual budget with minor tweaks, the rolling element is cosmetic. Build explicit decision triggers into the process, for example, if commodity costs move more than 5 percent from the driver assumption, the pricing committee convenes within 30 days.

The governance shift, reviewing driver assumptions before presenting numbers, is the change most CFOs underestimate. It is also the one with the highest resistance from business unit leaders who prefer to receive and dispute a final number rather than co-own the assumptions behind it.

One meaningful difference from the Unilever context: the company had the scale to build full-time FP&A capability in each major business unit. Smaller organisations with lean finance teams will need to centralise more of the driver modelling and accept less granularity at the product or geography level. That is a reasonable trade-off. A simple driver model that updates quarterly beats a detailed budget that nobody believes.

The method is not complicated. The discipline required to run it consistently, especially when business results are uncomfortable and the instinct is to reopen assumptions rather than accept what the model shows, is where most implementations stall.