Sales & Operations Planning (S&OP)

Sales & Operations Planning (S&OP) is the monthly cross-functional process that takes a cleaned demand forecast, balances it against supply capacity, financial budget and inventory targets, and produces a single signed-off plan that every function executes against for the next 18–24 months. It is a decision process, not a reporting cycle — the output is a set of approved trade-offs (build ahead, defer a launch, raise safety stock, hire shift labour, expedite a container) rather than a deck of charts.

The acronym is sometimes written 'S and OP' or 'S & OP'. Some organisations call the more strategic, finance-led variant Integrated Business Planning (IBP). The mechanics are similar; IBP simply pulls long-range scenarios, M&A and capex into the same cadence.

S&OP matters because every downstream warehouse metric — fill rate, days of supply, working capital, shift roster, replenishment workload — is a direct consequence of how cleanly the forecast flows into stock policy. A warehouse with no S&OP discipline lurches between stockouts and write-offs; a warehouse with disciplined S&OP runs a continuous loop: clean history, statistical baseline, judgemental override, signed-off plan, executable replenishment policy, measured accuracy, and feedback into next month's model.

The textbook S&OP cycle (Oliver Wight, APICS) runs five sequential steps each month. Most mature businesses complete the cycle in three to four weeks, with the executive review landing in the first week of the following month so the signed plan is in force for the period it covers.

The cadence is non-negotiable: an S&OP cycle that slips into the period it is supposed to govern has failed by definition. Mature teams publish the calendar 12 months ahead and treat the executive review like a board meeting.

Forecast accuracy is the single most important S&OP KPI because every downstream parameter (safety stock, ROP, EOQ, capacity plan) is calibrated against it. Three measures dominate.

• Track MAPE per product family at lag 1 (one month ahead) and lag 3 (three months, the typical lead time).
• Persistent bias (>5%) is a process problem, not a model problem — investigate sales overrides and promotional planning.
• WMAPE is fairer than MAPE when you mix fast and slow movers — MAPE punishes you for missing a single unit on a slow SKU.
• Report accuracy to the executive review every month. If it isn't measured, S&OP becomes a meeting, not a process.

An S&OP forecast that does not change stock policy is just a number on a slide. The forecast and its uncertainty should drive three executable parameters every cycle.

• Forecasting at the wrong level — too aggregated hides SKU mix, too granular is statistically meaningless. Family-level baseline, SKU-level disaggregation is the usual rule.
• Sales overrides accepted without governance — the consensus forecast becomes the sales target plus a wish.
• Safety stock set once and forgotten — every demand or lead-time shift erodes the achieved service level.
• Executive S&OP signed off but never reconciled to actual — no closed loop, no learning.
• Fill rate measured at order level only — line-level customer pain is invisible.
• S&OP run by planning in isolation — without finance and sales in the room, trade-offs cannot be made.
• No DG / regulated-product overlay — pharma and chemical S&OP plans that ignore controlled-stock policy create compliance gaps.

• Grocery — short shelf life forces tight forecast accuracy and daily replenishment cycles; promotional planning dominates the demand review.
• Fashion — high obsolescence risk, season-based planning with life-cycle curves, markdown plans baked into the supply review.
• Industrial spares — long tail, intermittent demand, Croston or bootstrap methods, service-level differentiation by criticality.
• Pharma & medical device — regulated supply continuity drives 98%+ service-level targets, batch genealogy and shelf-life constraints flow into stock policy.
• Subscription DTC — known customer base, deterministic forward demand, churn forecast more important than statistical baseline.