Operating Playbook: Demand Forecasting and Capacity Sequencing

Two functions, one future

Demand forecasting and capacity sequencing fail most often not because the forecast is wrong but because there are two of them. Sales builds a number to drive the commercial plan and, consciously or not, leans it optimistic. Operations builds a quieter number to protect against overbuild and leans it conservative. Each is internally reasonable, and they are reconciled, if at all, in a late meeting where the gap is split rather than resolved. The result is capacity that is sequenced against a future neither function actually believes: too much for the cautious number, too little for the ambitious one, and wrong for the compromise that gets built. The failure is structural. Two forecasts guarantee misaligned capacity no matter how good each one is.

This playbook runs demand and capacity as a single integrated cadence off one demand number that both functions sign. The forecast is built once, with sales and operations inputs reconciled inside the process rather than after it, and it becomes the only future the business plans against. Capacity additions, the new line, the extra shift, the third fulfillment node, are then sequenced against that shared number with explicit lead times, so the decision to add capacity is made early enough for the capacity to exist when the demand arrives. The discipline is not better forecasting in isolation. It is one forecast, jointly owned, driving a sequencing decision on a cadence.

Joint ownership is what makes the number durable. When sales and operations reconcile at the level of drivers, unit velocity, win rates, lead-time assumptions, the resulting forecast carries both functions' fingerprints, and neither can later disown it as the other team's optimism or the other team's caution. That shared authorship matters more than the last decimal of accuracy, because a forecast everyone signs gets acted on and a forecast one function distrusts gets quietly hedged against with a private second number. The integrated cadence exists precisely to kill that private second number, which is the thing that misaligns capacity in the first place.

Sequence capacity against one signed forecast

Build the demand-to-capacity sequencing table: the single signed demand number by quarter, the capacity it requires, the current capacity, and the sequenced action with its lead time. The worked example is a manufacturer whose base capacity is 100,000 units a quarter, where a new line adds 25,000 units but takes two quarters to commission. The table shows why the decision has to be made ahead of the gap.

The table makes the timing problem impossible to miss. The Q2 shortfall cannot be closed by building capacity in Q2, because the line takes two quarters to commission; the decision to cover Q2 had to be a bridge, and the decision to cover Q4 has to be made back in Q2. This is the entire reason demand and capacity must run off one number: the sequencing decision leads the demand by a lead time, so a forecast that arrives late, or that two functions still disagree on, arrives after the decision window has closed. On this book, missing the Q4 line-3 decision by one quarter forces a full quarter of outsourced production at roughly a 40 percent cost premium on 9,000 units, a self-inflicted cost that a shared, on-time forecast avoids entirely.

The table is also where the business decides its capacity buffer deliberately rather than by accident. Running capacity a few points ahead of the signed forecast, as in Q1 and Q3 here, is a priced insurance policy against forecast error and demand spikes; running it behind, as the bridges in Q2 and Q4 accept, trades a known premium for the flexibility of not committing fixed capacity too early. Making that buffer an explicit line in the sequencing review, rather than an emergent property of two functions' disagreement, is what lets leadership choose how much demand risk it wants to carry instead of discovering the answer after the quarter closes.

The cadence that keeps both functions aligned

• Produce one signed demand forecast per cycle, with sales and operations inputs reconciled inside the process, so there is a single future to plan against rather than two competing ones.
• Maintain the demand-to-capacity sequencing table as the shared operating artifact, showing the gap and the sequenced action with its lead time for every quarter in the horizon.
• Make each capacity decision at its lead-time deadline, not when the gap appears, so long-lead additions are committed the quarter or two before they are needed.
• Pre-define the bridging options, overtime, outsource, and inventory build, so short-notice gaps inside the lead time have a planned response rather than a scramble.
• Track forecast accuracy by quarter and feed it back into the reconciliation, so the shared number earns the trust that keeps both functions signing it.

What breaks demand-capacity alignment

• Running two forecasts. Sales optimistic, operations conservative, reconciled too late. Fix: one signed number, inputs reconciled inside the process.
• Deciding capacity when the gap appears. Long-lead additions cannot be built in the quarter they are needed. Fix: make each decision at its lead-time deadline.
• No planned bridge for short-notice gaps. Gaps inside the lead time force a costly scramble. Fix: pre-define overtime, outsource, and inventory-build options.
• Never checking forecast accuracy. An unaudited forecast loses the trust that keeps both functions signing it. Fix: track accuracy by quarter and feed it into reconciliation.
• Splitting the difference between the two numbers. The compromise fits neither function's real belief. Fix: reconcile the drivers, not the totals, to a number both can sign.

To run demand and capacity as one program

• There is one signed demand forecast per cycle, reconciled inside the process, not after it.
• The sequencing table shows the gap and the sequenced action with its lead time for every quarter.
• Each capacity decision is made at its lead-time deadline, ahead of the demand it covers.
• Bridging options for short-notice gaps are pre-defined, not improvised.
• Forecast accuracy is tracked by quarter and fed back into the shared reconciliation.