xP&A at Unilever: one forecast, three functions

September 4, 2026 · 5 min · Free, with full transcript

Can a business the size of Unilever really run one plan? The episode answers by tracing how finance, sales and supply chain each built their own forecast, a revenue plan, a demand plan and a volume plan that never reconciled until the quarter closed. The position taken is blunt: integrated planning is a governance outcome, not a technology purchase, and extended planning and analysis works only when one set of numbers is argued over in a shared model. Listeners get a way to test their own planning architecture: lay the revenue, demand and supply volume numbers side by side, check the gap, see where working capital and stranded stock leak, and decide who owns assumptions versus reconciliation.

Listen to the podcast

4 min

Chapters

Key takeaways

  • Pull your revenue, demand and supply forecasts and compare the volume number side by side; a gap beyond a few percent means the seams, not the tools, are broken.
  • Let the demand planner own demand assumptions and give finance ownership of reconciliation across the volume, revenue and cash plans.
  • Force disagreement into one shared model and one cadence so no team keeps a private version of the forecast after the meeting.
  • Treat the forecast as a binding contract; three unsigned plans are worth less than one number everyone commits to.
  • Budget years rather than quarters for an xP&A rebuild, with technology roughly a third of the work and process plus ego management the rest.
Read the full transcript

Host:Welcome to the Leaders Insights Podcast. Today's episode, How Unilever Rebuilt Its Planning Architecture Around XP and A. Everyone claims their planning is integrated. Unilever spent years proving it wasn't. What was actually broken?

Expert:Three planning engines running in three different rooms. Finance built a revenue plan. Sales built a demand plan. Supply chain built a volume plan. None of the numbers matched. Finance would forecast 14 billion euros in a region. Supply chain would build capacity for a different volume entirely. And nobody reconciled until the quarter closed.

Host:That sounds almost too dumb to be real at a company that size.

Expert:It's the opposite of dumb. Each silo was brilliantly optimized on its own. That's the trap. Sales optimized for hitting the number, so they sandbagged. Supply chain optimized for service levels, so they overbuilt inventory. Finance optimized for the board narrative. Every team was locally rational and collectively insane.

Host:So what does XP and A actually mean here? Because it sounds like another acronym invented to sell software.

Expert:Fair suspicion. XP and A is extended planning and analysis. The idea that finance's forecasting discipline gets stretched across every operational plan. Not just the money one. The point isn't a new tool. The point is one set of numbers that all three functions are forced to argue over in the same room, at the same cadence.

Host:Forced to argue. That's the part people skip. Everyone wants alignment without the fight.

Expert:Right, and that's why most integration projects fail. Alignment isn't a technology outcome. It's a governance outcome. Unilever didn't just plug systems together. They rebuilt who owns the assumptions. When sales says demand is down 5%, finance and supply chain now have to accept or challenge that in a shared model. You can't quietly keep your own version anymore.

Host:But someone has to own the master number. Who wins that fight? Finance always?

Expert:Finance owns the reconciliation, not the assumption. That's the subtle bit. The demand planner still owns demand. They know the promotions, the shelf data. Finance's job is to hold the seams. To make sure the volume plan, the revenue plan, and the cash plan are describing the same reality.

Host:Referee, not dictator. Give me a number. What does this mess actually cost a business before they fix it?

Expert:Working capital is where it bleeds. When forecasts disconnect, you carry inventory you don't need against demand that never shows. For a consumer goods giant, a few points of forecast error across 60 billion euros in sales is hundreds of millions in stranded stock. Gartner, and yes, they sell to this exact buyer, so discount accordingly, has pegged forecast accuracy gaps at 3 to 5 percentage points of revenue in avoidable cost. Even at the low end, that's real money.

Host:So the durable lesson isn't by the platform. What is it?

Expert:The lesson the good operators already know. A forecast isn't a document, it's a contract. If sales, supply chain, and finance produce three forecasts, you don't have three plans. You have zero, because none of them is binding. One number that everyone signs is worth more than three sophisticated ones nobody trusts.

Host:That's a nice line. But doesn't one number just paper over genuine disagreement? Sometimes the functions should see the world differently.

Expert:They should. Before the meeting, disagreement is the input. What you can't tolerate is disagreement surviving the meeting as parallel spreadsheets. Get the assumption fiercely, then commit to one output. The failure mode isn't conflict. It's unresolved conflict that each team keeps privately in their own model and acts on anyway.

Host:How long did Unilever's rebuild take? Because I suspect the honest answer scares people off.

Expert:Years, not quarters. And the technology was maybe a third of it. The rest was process, and frankly, ego management. Getting powerful function heads to give up their private version of the truth. Someone selling you a six-month XP and A transformation is selling you a dashboard, not a rebuild.

Host:Give me the one thing a listener can do Monday morning.

Expert:Pull the three forecasts you already have. Revenue, demand, and supply is he and lay the volume number side by side. If they don't match to within a few percent, you don't have a forecasting problem. You have three companies pretending to be one. Fix the seams before you touch the tools.

Host:That's it. Go to UnileverDecisionTools or live at MBA-training.com.

Read and go deeper