Sales has a number. Operations has a different number. Finance has a third, which matches neither. Everyone knows this and everyone has stopped finding it strange, which is itself the problem.

Three Numbers, Three Purposes

The disagreement isn’t usually incompetence. It’s that each number is built for a different job.
Sales forecasts revenue, in currency, by customer and region, aimed at a target. Operations plans units, by SKU and by week, constrained by capacity. Finance builds a budget that has to be defensible to the board. These are three genuinely different questions, and answering all of them with one spreadsheet was never going to work.
The trouble starts when nobody converts between them. Revenue by region doesn’t tell a planner how many of SKU 4471 to build in week 32. Someone has to translate, and if that translation is informal — a planner making assumptions about mix — the gap between the numbers becomes invisible until it causes a shortage.

The Incentive Problem

Sales is measured on hitting targets, which makes optimism rational — forecasting high protects against being caught short and rarely carries a personal cost. Operations is measured on service and cost, which makes conservative capacity commitments rational. Both behave sensibly within their own incentives and produce numbers that can’t both be right.
No process fixes this if the incentives stay untouched. The most effective single change most manufacturers can make is to measure sales on forecast accuracy alongside revenue, and to publish the accuracy by region. Not as punishment — as feedback. Forecasting improves fast when people can see their own track record.

Making the Translation Explicit

Build a documented conversion from the revenue forecast to a unit forecast by SKU, using historical mix, and let everyone see the assumptions. When operations disagrees, the disagreement is now about a specific assumption — that this customer will keep buying the same mix — rather than a general sense that sales is over-promising.

Then agree one number. Not three reconciled numbers. One consensus forecast that sales, operations, and finance all plan against, with the differences resolved in a meeting rather than carried forward silently. This is the core of what S&OP is meant to deliver, and where most S&OP implementations quietly fail: the meeting happens, the numbers are presented, and everyone goes back to their own version.

Keeping it Honest

Record the consensus forecast and compare it to actuals every month, by product family and region. Look at bias direction, not just error size — consistently high or consistently low is a fixable pattern, while random error is just uncertainty. Six months of this data changes the conversation more than any amount of process redesign.
It also changes who gets listened to. When a regional manager’s forecasts have been demonstrably accurate for two quarters, their judgement on an unusual month carries weight it wouldn’t otherwise. That’s the quiet benefit of measuring: it lets you distinguish informed local knowledge from optimism, which is currently a distinction most planning meetings cannot make.
ticktick.ai maintains one demand plan feeding both procurement and production, and tracks forecast bias by source so the pattern is visible rather than argued about.