System Thinking · Supply Chain

Planning for a future you can’t see

Mohit Saini · 12 Aug 2025 · 4 min read

Planning for a future you can’t see

Every year, around the same time, most companies go through the same ritual. Sales teams forecast next year’s demand. Finance turns the forecast into a budget. The budget becomes targets for every department, and the targets become the basis for appraisals and incentives.

By the end of the first quarter, reality has usually wandered off from the plan. The rest of the year is spent explaining the variance.

Why do we keep planning as if we can see the future, when experience keeps telling us we can’t?

Experts are good at explaining, not predicting

Think of any big surprise of the last twenty years: a commodity price that collapsed when everyone expected it to soar, a product launch that was supposed to change an industry and quietly faded, an expansion boom that turned into a debt crisis a few years later.

Before each event, there were confident forecasts, backed by detailed studies. After each event, there were equally confident explanations of why it happened. The explanations always sound obvious. That is the trap: because we can explain the past so well, we believe we can predict the future.

The evidence says otherwise. The psychologist Philip Tetlock tracked thousands of predictions by experts over two decades. For long-term forecasts, the experts did little better than chance.

You can run your own test. Ask the founder of any company that has been successful for twenty years whether things went according to the original long-term plan. Or read business magazines from twenty years ago and check their predictions about today.

Why markets are so hard to forecast

Most demand forecasts are built on a simple assumption: that the factors driving demand act independently and proportionately. But markets are full of players who react to each other:

  • A product becomes scarce, so dealers start ordering extra, which makes it scarcer.
  • A company launches a scheme, competitors respond, and the next month’s sales fall as the channel works through the extra stock.
  • A small change can snowball into a big effect, while a huge marketing campaign can sometimes change nothing at all.

These feedback loops make markets non-linear. We can understand the patterns once they have formed, but it is very hard to say in advance what number we will sell in a given month.

There is one useful observation, though. Forecasts get better as we get closer to the event. Tomorrow’s weather forecast is quite reliable; next month’s is not. Beyond a certain horizon, a forecast is closer to a guess than a plan.

Targets make the problem worse

When a long-range forecast is converted into fixed targets, two more things happen.

First, the organisation becomes rigid. Everyone is committed to one scenario. When the market changes, people keep chasing the original numbers instead of responding to what they see.

Second, the measures get distorted. Once sales volume becomes a target, it stops being a clean measure of demand. People shift orders between months, push stock at month end, or hold back to avoid a higher target next year. This is Goodhart’s law at work: when a measure becomes a target, it ceases to be a good measure.

The result is an organisation that is slow to sense change and slow to respond, precisely when speed matters most.

The way out: react faster, experiment smaller

If we accept that the future can’t be known, there are two practical ways to manage.

1. Build the ability to react quickly

Instead of committing resources early based on a forecast, hold back commitments until the uncertainty reduces, and be fast enough to respond when it does. This means cutting lead times everywhere: in production, distribution, new product development, capital projects, and even management decisions.

Some of the best examples are in supply chains:

  • Fast-fashion retailers that design and deliver a new style within weeks of seeing what sells, instead of betting on trend forecasts made nine months before the season.
  • Distribution systems that keep most stock at a central warehouse and send it to locations only as actual daily sales pull it, so that stock flows to where demand turns out to be, not where it was forecast to be.

Neither tries to predict better. Both are built to need less prediction.

2. Test ideas in small, safe steps

The second mode is for creating the future rather than reacting to it: new products, new markets, new business models. Since we can’t be sure which ideas will work, test them at a scale where failure is affordable, and be ready to scale up fast when an idea succeeds.

The same applies to big investments. Instead of one large capacity expansion based on a five-year forecast, break it into phases. Trigger each phase by the results of the previous one. If the boom lasts, you keep building. If it doesn’t, you haven’t bet the company on a guess.

Stop trying to see further. Learn to react faster.

In short

Forecasts are not useless. They help us think about scenarios. The danger lies in treating a long-range forecast as a commitment, and building the whole organisation around it.

Accepting “we don’t know” is uncomfortable. But it leads to better questions: How fast can we respond? How small can we make our bets? Interestingly, “live in the present” is not just good advice for personal well-being. It is a practical way to run a company in an uncertain world.

Related reading

Mohit Saini

Strategy, supply chain & sales transformation consultant. About · Book a discussion