The Difference Between Planning for Stability and Planning for Reality

Years ago, I sat in on an S&OP meeting at a specialty chemicals manufacturer where the demand planner presented a number: 42,000 units for the month. Clean, confident, single line on a slide. The room nodded. Production scheduled to it. Procurement bought raw material against it.

Actual demand came in at 51,000. Not because anyone had made a mistake. Because the number the plan hung on was never meant to be exact. It was a midpoint. And a midpoint is a plan for the one outcome that’s least likely to happen exactly as stated.

Ask any planner in that room and they’ll tell you demand is a range. But the system doesn’t run on what planners believe. It runs on the number they’re forced to enter a single field. That gap between what experienced planners know and what the planning math uses is the difference between planning for stability and planning for reality.

Planning For Stability Isn’t Wrong. It’s Incomplete.

Most planning systems were built around a reasonable assumption: that the world holds still long enough for a single forecast, a quoted lead time, and a rated capacity to be good enough. For a long stretch of time, that assumption mostly held.

It shows up everywhere. Demand gets reduced to one number per period, even though every planner knows the real question is “what’s the range, and what’s driving each end of it.” Lead times get entered as the quoted contract figure, not the distribution of what’s actually shown up over the last eight quarters. Capacity gets modeled at rated throughput, not the effective throughput a line holds after changeovers and campaign length.

None of these simplifications were arbitrary. They made the math tractable when compute was expensive. The problem is the operating environment they were built for — steady demand, dependable suppliers, predictable throughput — isn’t the one most manufacturers run in anymore. Supplier concentration, weather-driven variability, and demand volatility have moved from occasional to structural. The math didn’t move with it.

The Tell Is In The Safety Stock.

Look at how safety stock gets set. In most operations I’ve reviewed, it’s a static number — a fixed number of days of supply, applied uniformly, revisited once a year if at all.

I’ve sat with planning teams who could tell me, item by item, which SKUs actually needed that buffer, because they knew from memory which suppliers ran late and which items had volatile yield. That knowledge was accurate. It just never made it into the formula. The planners compensated by hand, every cycle, quietly overriding a number that was never built to reflect the risk they could see plainly in front of them.

That’s the clearest evidence of the divide. Planning for stability produces one number and asks people to manage the gap between it and reality. Planning for reality tries to put the gap into the number itself.

What Planning For Reality Looks Like

This isn’t a call to abandon forecasts or lead times. It’s a call to stop treating their midpoints as the whole answer.

Planning for reality means carrying demand as a range with the drivers behind each end made explicit — not “demand could run higher,” but which customers and regions are pushing that upper bound. It means modeling lead time as a distribution built from actual delivery history, not the contract number. It means capacity plans that reflect what a line actually produces across a realistic run length, including the drop-off every scheduler already works around by hand.

The output isn’t a more complicated plan. It’s a plan that tells you more — where it’s tight, where it has room, and what it costs to buy more room where it matters. Running plans this way at weekly or daily speed wasn’t realistic by hand, and it wasn’t realistic in a lot of planning software until fairly recently. What’s changed is that carrying ranges through an optimization, and explaining plainly why the plan landed where it did, is now fast enough to be routine rather than a special project.

The Question I’d Ask Your Planning Team

Pull up your current plan and ask how many of its numbers are single points, and how many reflect a range your team already knows exists. For most organizations, the range lives in planners’ heads while the single point lives in the system — reconciled only through manual overrides that never make it back into the model.

That gap is closeable. Not by hoping for calmer conditions, and not by hiring more people to manage the overrides. By building plans on ranges instead of averages, and letting the plan tell you what those ranges cost to cover.

The forecast was never the problem. Planning as if the forecast were a certainty was.

See how this thinking comes together across constraint modeling, tacit knowledge, and planning for reality on the Designing Resilience Into the Plan hub.