Ask five people what the sales forecast means, and you will get five different answers. This is one of the most common sources of organisational friction, and one of the least discussed because it is so embedded in how businesses operate that it becomes invisible. 

What operations needs

In principle, the ideal demand forecast is a 50/50 number. Equal probability of being too high or too low. That is what you need to plan inventory, capacity, and purchasing sensibly. It does not need to be perfectly accurate, as there are tools to deal with variability. It just needs to carry no systematic lean in either direction. Operations do not need optimism or caution. It needs the best estimate of what is actually going to happen. 

In theory, this is perfectly achievable. 

Enter real life

In reality, the number is shaped long before anyone opens a spreadsheet. Nobody forecasts their own business loss. Which business worth its salt will make a plan to lose? We may predict a decline in current circumstances, but we rarely plan one. We fight. We make new plans, commit new budget lines, and then, of course, forecast the sales recovery they are supposed to deliver. At the level of individual initiatives and marketing activities, the same dynamic plays out: once a number has been used to justify a budget, stepping away from it becomes institutionally very difficult, even when the original assumptions no longer hold. 

Meanwhile, sales need a number they can meet, because incentives are tied to it. An achievable target for each product line, which tends to err on the conservative side. Finance wants a number that satisfies the board, something with the flavour of cautious optimism. These pressures are forced into a single figure through a consensus process that is really a negotiation. The result is a number shaped by science, previous commitments, politics, and incentives. Not necessarily what anyone privately believes will happen. 

Most organisations develop a persistent lean in one direction or the other, depending on culture, power dynamics, and the perceived cost of being wrong. These biases can become embedded and self-reinforcing, to the point where correcting the forecast carries a higher perceived organisational cost than continuing to miss it every month. 

These biases come with a heavy price tag. Misallocated resources, supply shortages, excess product. Over time, they also erode trust in the forecast itself. When people stop believing the number, they stop following it. Instead, they introduce their own overrides, adjusting the plan based on what they think will really happen. I call it “forecasting the forecast”, and it renders the original exercise almost useless. 

If any of this sounds familiar, you are in good company. 

At SME scale, where the same person often wears more than one of these hats, the tension does not disappear. It just plays out inside one person’s head rather than across departments. That makes it harder to see and harder to fix. 

The tools will not save you

There is an extensive industry dedicated to improving forecast accuracy. Better algorithms, machine learning, demand sensing, all dedicated to the optimisation of SKU-level forecast to the deepest detail. These tools have their place. But if the much larger foundation feeding the system has already been shaped by budget politics, incentive structures, and unspoken optimism (or pessimism) before a single algorithm touches it, then improving the last 1% of statistical accuracy can only take you so far. 

What good looks like, honestly 

“Do we plan the sales or do we sell the plan?”. I do not think there is a magic answer to these questions. 

The conflict between aspiration and reality is inherent in how organisations work. It does not get solved. It gets managed. 

Some organisations separate the numbers (the demand signal, the financial commitment, the sales target) and name them for what they are. Others maintain a strict “one number” policy. What is important is to create a space where the gap between those concepts can be discussed without someone losing face. To treat revision as information, not failure. 

A mature S&OP process does exactly this. Not through elaborate governance, but through the discipline of confronting reality collaboratively. It recognises that different parts of the business have different incentives, and that pretending otherwise produces worse outcomes than acknowledging it. 

The organisations that forecast well are not necessarily those with the best tools or the most sophisticated processes. They are the ones with high empathy between functions, and the ones that have learned to be honest about what they know, what they do not know, and what they are quietly pretending not to know.