S&OP That Produces WIP Instead of a Plan
Halden Industrial builds configurable conveyor systems to order in three product families. Long-lead drive motors take 18 weeks; frames are fabricated in-house in 4 weeks. Monthly S&OP has become a forecast-approval ritual: sales submits an optimistic number, operations habitually discounts it 15%, and the general manager splits the difference. Family-level forecast error has averaged 58% MAPE for eight quarters. Because planners are measured on 'plan attainment,' they release orders early. Work-in-process has doubled to $9.3M and on-time delivery sits at 84%. A customer representing $4M of annual revenue is threatening to move its program.
1. Diagnose why the S&OP process is producing these symptoms. Which causes are structural and which are behavioral?
Read the worked answer
Structural: the company is forecasting at the wrong level. Finished configurations of a make-to-order product cannot be forecast; the forecastable elements are the options and modules - motor type, drive, frame length. Planning should happen at that level with final configuration postponed to the order. Separately, an 18-week motor lead time exceeds any useful family forecast horizon, so the motor needs a different mechanism entirely: capacity reservation or a buffer, not a forecast.
Metric-induced: measuring planners on plan attainment rewards releasing early, which is precisely what generates $9.3M of WIP and still leaves OTIF at 84%. The measure and the goal are in direct conflict.
Behavioral: habitual 15% discounting means the sales number carries no information and nobody owns it; bias is never measured, only error. The GM splitting the difference proves the pre-meeting is failing - disagreement is being resolved in the executive room instead of before it. Together these turn S&OP into a negotiation over a number nobody believes.
2. What do you change in the next 90 days, and what do you measure instead?
Read the worked answer
Structure
- One consensus demand plan, signed by sales and operations, expressed at module/option level.
- Report bias (signed) alongside MAPE. A persistent +15% bias is a fixable process defect; 58% error alone is not actionable.
- Pre-S&OP with ops and finance five days before the executive meeting; the executive meeting only resolves documented gaps.
Flow
- Put a WIP cap on each work center and release frames only to a written policy (confirmed orders plus a defined buffer), not to a plan number.
- Replace plan attainment with schedule adherence, OTIF, WIP turns, and past-due backlog.
Motors
- Reserve rolling six-month capacity with the motor supplier or hold a strategic buffer of the top three motor variants. This is a real trade: a few hundred thousand dollars of buffer against a $4M program at risk - the arithmetic favors the buffer, but test it against the actual probability of loss rather than assuming it.
Track plan stability (frozen-period adherence) too. If the plan is re-cut weekly, no supplier or planner can be held accountable for it.
3. The $4M customer demands a six-week pull-in that requires taking a motor already reserved for another customer's order. How do you decide?
Read the worked answer
Decide on contribution margin at risk on both sides, not on revenue, and include contractual penalties and switching risk.
Work the options in cost order before cannibalizing:
- Can the unit ship with an alternate motor variant the customer will approve? Often yes, with a small performance delta.
- Can the supplier expedite or pull from a distributor at a premium? Compare the premium to the penalty and the margin at risk.
- Can the order be split - partial shipment now, balance on the original date?
- Only then compare the two customers: margin at risk, penalty exposure, lifetime value, and how recoverable each relationship is.
Whatever the outcome, make it an escalated, documented decision with a stated rule - the worst outcome is a planner quietly reshuffling motors, because that destroys the credibility of every future promise. Then fix the cause: reservation policy, buffer sizing, and quoting lead times that reflect motor availability rather than frame capacity.