How Unilever rebuilt its forecasting on drivers, not budgets

August 27, 2026 · 5 min · Free, with full transcript

Unilever's finance function spent years trapped in a cycle of annual budgets that were outdated before the ink dried. Its move to driver-based, rolling forecasts offers a detailed blueprint for CFOs who want forecasts that actually inform decisions.

Listen to the podcast

4 min

Read the full transcript

Welcome to the Leaders Insights Podcast.

Today's episode, how Unilever rebuilt its forecasting on drivers, not budgets.

Every finance team on earth builds an annual budget, then spends 11 months explaining

why it was wrong.

Why do we keep doing it?

Because it feels safe.

You lock a number in November, and by March the currencies moved.

A competitor's cut price, and your beautiful spreadsheet is a museum piece.

But the ritual survives because someone gets to say, we hit budget.

Even when hitting budget meant ignoring reality.

Unilever finally admitted the emperor had no clothes.

What did they actually change?

They stopped forecasting the budget line and started forecasting the drivers underneath

it.

A driver is the real world thing that moves the number.

Volume sold, price per unit, raw material cost, the marketing spend behind a launch.

Instead of guessing, revenue will be $62 billion, U-model will sell this many units at this

price with this input cost, and revenue falls out of that.

That sounds like more work, not less.

You've replaced one number with 50.

You have, eeeeeeee.

But you've replaced 50 guesses with maybe 8 or 10 that genuinely move the business.

That's the discipline nobody talks about.

Driver didn't model everything.

They found the handful of drivers that explained most of the variants.

Palm oil, sugar, packaging, promotional intensity in a few key markets, and let the rest ride.

If a driver doesn't change your decision, stop forecasting it to two decimal places.

Give me the mechanics.

How does driver-based actually beat the old way when the numbers land?

Take input costs.

In the old model, if palm oil spikes 30%, you find out at month end when margins already

bleeding, and finance writes a variance report—a note explaining the gap between plan and actual.

In a driver model, the price feed updates the forecast the same week.

And you can see three months out that gross margin's about to fall two points.

That's the difference between a warning and an autopsy.

But someone's still typing assumptions into a model.

Garbage in, garbage out.

Why trust the new garbage more than the old?

Because the assumptions are now visible and challengeable.

When your forecast says, revenues up 4%, I can ask, is that volume or price?

And you have to answer.

In a budget, 4% is a black box you negotiated in a room.

Drivers force honesty.

And you can pressure test them.

Anna plan in pigment.

The planning software vendors pushing this hard.

Both love quoting that driver-based shops run scenarios in hours instead of weeks.

Take that with salt—they're selling the tool.

But the underlying point holds.

If your model is built on levers, you can pull a lever and see what happens.

Unilever also went to rolling forecasts.

Explain that without the jargon, and tell me why it's not just extra meetings.

A rolling forecast means you always look 18 months ahead, refreshed every quarter.

So the horizon moves with you instead of shrinking as the year runs out.

The old annual budget gets myopic by October.

After forecasting eight weeks and calling it strategy, rolling keeps the runway constant.

And no, it's not more meetings if you kill the old ones.

Unilever's whole point was to retire the annual budget theater, not stack a new process

on top of it.

That's the catch, though.

Most companies add the rolling forecast and keep the budget.

Now they've got two.

And that's why most transformations fail.

You cannot run both.

If the budget still controls bonuses and the forecast is just for information, people game

the budget and ignore the forecast.

Unilever had to move the accountability—the incentives, the board conversation—on to

the rolling view.

Otherwise you've built a second dashboard nobody steers by.

So for the CFO listening who's trapped in the November budget cycle right now, where

do they start Monday?

Pick your top three drivers—the three variables that, if they move, actually change what you

decide.

Build a one-page forecast off those, update it monthly, and put it next to your budget.

Within a quarter, you'll see which document predicted reality.

And I promise it won't be the budget.

Then you've got the evidence to kill the thing.

Three drivers and a one-page page.

That's the homework.

Thanks.

That's it.

The CFO decision tools are live at MBA-training.com.

Read and go deeper