S&OP: cycle and documents
S&OP (sales and operations planning) is a monthly cycle in which commerce, production, purchasing and finance reconcile the demand plan, the supply plan and the money into one decision. It resolves the conflict where sales want more, the factory can do less and finance gives less, and records which single plan is adopted.
Why this document
Without a shared cycle each department has its own plan: commerce believes in growth, the factory counts capacity, finance limits cash, and the mismatch surfaces when it is already late. In S&OP discrepancies are exposed once a month while they can be fixed: shift an order, reallocate capacity, release money. The owner or board adopts one plan and owns the compromise. So what matters is not presentations but decisions: what changed, who is responsible and by what date.
At a glance
- What is this document and why is it needed?
- S&OP (sales and operations planning) is a monthly cycle in which commerce, production, purchasing and finance reconcile the demand plan, the supply plan and the money into one decision. It resolves the conflict where sales want more, the factory can do less and finance gives less, and records which single plan is adopted.
- Who owns the document and how often is it rebuilt?
- Brand: the commercial director runs the cycle, a planner prepares the documents, the finance director and the operations director take part; the owner or board approves. Retailer: the sales and buying director, finance, logistics; a shorter cycle closer to categories.
- What mistakes are the most common?
- S&OP becomes a report with no decisions → meetings do not change the plan → each cycle ends with a decision list with owners. The commercial forecast diverges from the purchase plan → shortage or overbuying → a single consensus forecast. Bias is ignored → the forecast is always above actual → compute bias and correct.
For whom
- Brand
- A brand reconciles the demand plan (wholesale, retail, online) with factory capacity, fabric lead times and the budget: decisions on fabric booking, early and late orders, reorders. Long lead times make the cycle multi-level: annual, seasonal, monthly.
- Retail
- A retailer reconciles the category sales plan with OTB, deliveries and the warehouse; the cycle is usually weekly and monthly; it confirms brand delivery dates and warehouse capacity and decides on reallocating budget between categories.
What it consists of
The cycle document pack
- 1. Sales review: actual against plan, forecast accuracy (WAPE), bias.
- 2. Demand plan: a consensus forecast of commerce, marketing and channels.
- 3. Supply plan: orders, capacity, dates, warehouse, constraints.
- 4. Financial roll-up: revenue, margin, cash, stock; scenarios.
- 5. Decisions and risks: what is approved, who owns it, by when.
Cycle calendar
- Weeks of the month: collect actuals → demand review → supply review → pre-S&OP → final decision.
- The window rule: decisions for the next four weeks are frozen and changed only with the consent of the demand and supply owners; beyond that the window is flexible. Process indicators: the share of decisions done on time, the number of escalations, the number of revisions of the frozen window, the share of the forecast revised after a trigger fired.
- Preparation: the process owner (usually the finance or commercial director), an analyst, representatives of commerce, marketing, purchasing, logistics and finance; materials are sent a day before the meeting, the meeting discusses only divergences and decisions, not a retelling of the tables.
Measures
- Forecast WAPE and bias; tracking signal; purchase plan adherence (plan vs actual); supplier OTIF; weeks of supply.
The brand “demand — supply — money” summary (Model B), the decision log, accuracy at SKU level
- Brand: Model B by quarter (units; purchases of RUB 97.546 million a year ÷ 1,200 = 81,288 units, opening stock 43,333): quarter 1: demand 16,541, supply 22,761, closing stock 49,553, closing cash RUB 6.275 million; quarter 2: demand 18,044, supply 21,135, closing stock 52,644, closing cash −RUB 3.5 million; quarter 3: demand 19,548, supply 19,509, closing stock 52,605, closing cash RUB 5.009 million; quarter 4: demand 21,052, supply 17,883, closing stock 49,436, closing cash RUB 14.351 million. Demand here is stock consumption (sold and written-off units, cost ÷ 1,200) by the seasonal shares of cash receipts, so year-end stock equals the Model B balance sheet; negative cash in the weakest quarter shows when a funding line is needed.
- The S&OP decision log: date; question; options; decision; owner; due date; cash impact (RUB thousand); status (accepted, done, revised). Weekly agenda: week 1 collect actuals and WAPE by category; week 2 demand review and consensus of commerce, marketing and channels; week 3 supply and capacity review; week 4 financial summary and decision, minutes within the day.
- Accuracy at the level where purchases are decided: over 12 “item × month” cells WAPE is 22.6%, against WAPE of 10% on the six-month aggregate: the aggregate hides errors that cancel each other. For decisions on sizes and items compute WAPE by purchase cells and show both figures.
Brand and retailer: what is computed differently
- Brand: a brand matches the demand plan (wholesale, site, marketplaces) with factory capacity, fabric lead times and the budget: decisions on fabric orders, early and late orders and reorders; the cycle is multi-level, annual, seasonal and monthly.
- Retailer: a retailer matches the category sales plan with OTB, brand deliveries and warehouse capacity; the cycle is weekly and monthly, with decisions on reallocating the budget between categories and on re-supplies.
Brand: how the document differs and what to add
- Brand: brand S&OP links demand (preorders, sell-out, site), production (MOQ, factory lead times) and cash (advances, deferrals); the cycle runs monthly.
- Brand: the brand horizon is longer than retail: production decisions are taken 6 to 9 months ahead, so the cycle includes a season ahead and updates by preorder.
- Brand: brand demand is distributed by channel (wholesale, site, marketplace); each has its own forecast and source: order book, traffic, search positions.
- Brand: supply decisions: postpone, speed up (air instead of sea), split into batches; each decision has a cost and is recorded in the log.
- Brand: cash in the cycle: factory advance, import payments, wholesale receivables; for each S&OP scenario the brand sees the cash minimum by quarter.
- Brand: the brand forecast is compared with actual partner sell-out, not shipments: otherwise the cycle rests on stock shifting.
Retailer: how the document differs and what to add
- Retailer: retailer S&OP links sales by category and store, purchasing (OTB), stock and cash; the cycle is weekly or biweekly in season.
- Retailer: the retailer forecast is built on traffic, conversion, ticket and the event calendar; units by SKU and size are secondary, the group level primary.
- Retailer: retailer triggers: sell-through below plan, stock age, mall traffic; each trigger leads to a decision on purchasing, price or transfer.
- Retailer: assortment decisions (delisting and listing) and pricing decisions are recorded in the decision log with a date, owner and expected effect.
- Retailer: the retailer compares forecast and actual at week and product-group level; accuracy at SKU level is shown separately as reference.
- Retailer: retailer cash in the cycle: supplier payments, rent, payroll and the seasonal peak; a purchasing decision is checked against the cash minimum.
- Retailer: the report on the implementation of last cycle decisions is the first item of the meeting: what was not done without a reason does not carry into the new cycle.
How to read the result and which decisions follow
- Forecast bias (BIAS 5.4%) is above a few percent and the tracking signal is beyond ±4 (in the example 3.2: the threshold is not crossed, the bias is not yet proven): the team over- or under-forecasts systematically; an independent review and model correction before the next cycle. A bias alone without the signal crossing the threshold is a reason to watch, not to change the model.
- Stock by quarter grows while cash falls to the minimum: shift supply or the reserve order into the next quarter; such a decision is taken in the S&OP cycle, not in purchasing.
- Channel demand grows while production cannot keep up: decide whether to speed up delivery (cost) or lose sales; the decision with numbers goes in the log.
Control checks
- Forecast 100 + 110 + 120 + 130 + 125 + 140 = 725, actual 96 + 118 + 103 + 138 + 112 + 121 = 688; sum of errors 37, bias 5.4% = 37 ÷ 688; WAPE 10% = Σ|errors| 69 ÷ 688; MAD 11.5, tracking signal 3.2 = 37 ÷ MAD.
- The demand, supply and cash figures in the three functions agree; one plan version is in force after the meeting.
- Accuracy is computed at the level at which the decision is made (group, not SKU) and by one rule through the year.
- Typical traps: the forecast is edited to fit the plan, a decision without an owner and date, cash not shown in the cycle, shipment actual taken for sales actual.
Where the data comes from
| Data source | Fields |
|---|---|
| Sales and forecast | Actual and forecast by week and category for 6-12 months; forecast versions. |
| Production and procurement | Orders, statuses, dates, OTIF, factory capacity. |
| Finance and warehouse | Budget, cash flow, stock, warehouse capacity. |
How to build it step by step
- Step 1. Close the period and collect actual sales and deliveries.
- Step 2. Compute forecast accuracy: WAPE, bias, tracking signal; flag categories with systematic bias.
- Step 3. Build the consensus demand forecast: commerce, marketing, channels; record assumptions and promos.
- Step 4. Match the supply plan to demand: what can be made and delivered, and when; find shortages and excesses.
- Step 5. Turn gaps into money: revenue, margin, stock, cash flow; prepare 2-3 scenarios.
- Step 6. Pre-S&OP meeting: settle everything solvable at function level, escalate the rest.
- Step 7. Executive meeting: choose a scenario, approve changes to the purchase plan and budget.
- Step 8. Distribute decisions with owners and dates; check execution in the next cycle.
Formulas
| Measure | Formulas | Note |
|---|---|---|
| Forecast error | The definition of WAPE and bias: the term “Forecast error (WAPE) and bias” and the “Forecast accuracy by week” calculation; in S&OP it is taken by month | A weighted error: not distorted by small items. |
| Bias | Bias = Σ(Forecast − Actual) ÷ Σ Actual × 100 | A positive value means systematic overforecasting and the risk of excess stock. |
| Tracking signal | TS = Σ(Forecast − Actual) ÷ MAD, where MAD is the mean |error| | Values beyond ±4 (as a hint, not a norm) signal systematic bias. |
| Demand-supply balance | Balance = Stock + Supply − Demand | A negative balance means shortage, a positive one means surplus. |
Formulas in notation
- Forecast bias
BIAS = Σ (F_t − A_t) / Σ A_t
F_t is forecast, A_t is actual; a positive value means systematic over-forecasting.- Accuracy
WAPE = Σ |F_t − A_t| / Σ A_t
Weighted absolute error; accuracy at SKU level is lower than for a group, so it is computed at the required level.- Stock by quarter
Stock_q = Stock_(q−1) + Supply_q − Demand_q
Units; conversion to money at unit cost gives the cash need for stock.
Worked example
Forecast accuracy over six months (thousand units) (the forecast accuracy is an independent example; the brand summary is built on Model B inputs)
| Month | Forecast | Actual | Error | Absolute error |
|---|---|---|---|---|
| 1 | 100 | 96 | 4 | 4 |
| 2 | 110 | 118 | −8 | 8 |
| 3 | 120 | 103 | 17 | 17 |
| 4 | 130 | 138 | −8 | 8 |
| 5 | 125 | 112 | 13 | 13 |
| 6 | 140 | 121 | 19 | 19 |
| Total | 725 | 688 | 37 | 69 |
WAPE 10%, bias 5.4% (the forecast is above actual in 4 of 6 months), tracking signal 3.2: it is within the ±4 threshold, so a systematic bias is possible but not proven; keep watching the signal and revise the forecast model if it goes beyond ±4. A positive bias, if confirmed, means inflated orders and growing stock: in the S&OP cycle the forecast is adjusted down and reorders are reviewed. A brand will compare it with factory capacity and cash, a retailer with OTB and warehouse capacity.
Figures
- Demand
- Supply
- Closing stock
Data as a table
| Measure | Demand | Supply | Closing stock |
|---|---|---|---|
| Q1 | 16.5k | 22.8k | 49.6k |
| Q2 | 18k | 21.1k | 52.6k |
| Q3 | 19.5k | 19.5k | 52.6k |
| Q4 | 21.1k | 17.9k | 49.4k |
Brand demand, supply and stock by quarter. Supply runs ahead of demand in the first half of the year and stock grows; then demand catches up and cash passes its minimum: the S&OP decision is how to shift supply.
- Forecast
- Actual
Data as a table
| Measure | Forecast | Actual |
|---|---|---|
| 1 | 100 | 96 |
| 2 | 110 | 118 |
| 3 | 120 | 103 |
| 4 | 130 | 138 |
| 5 | 125 | 112 |
| 6 | 140 | 121 |
Forecast and actual by month. The forecast is now above and now below actual: bias (BIAS) shows systematic error and accuracy (WAPE) shows error size; together they decide whether the forecast can be trusted.
Specifics by category
- Apparel
- Long lead times and seasonality: the main cycle is seasonal and refined monthly; fabric and early-order decisions are made 6-9 months before sales.
- Footwear
- The longest development cycle and the toughest MOQs; last and tooling decisions come earlier; a size-level forecast is mandatory; errors cost more because of the narrow core.
- Accessories
- Short cycles for basics and long ones for tooling and metal; a unit forecast with no sizes is simpler; gift peaks are handled separately in the cycle.
- Kids
- The cycle is tied to the school and winter season; the forecast depends on demographics and prices; document and certification requirements go into the calendar early.
- Men
- The forecast is steadier with a lower WAPE; the cycle can be simplified; reorders are more frequent thanks to basics.
- Women
- The most volatile demand: forecast error is higher, scenarios and early/late orders are needed; the cycle includes capsules and drops.
Common mistakes
- S&OP becomes a report with no decisions → meetings do not change the plan → each cycle ends with a decision list with owners.
- The commercial forecast diverges from the purchase plan → shortage or overbuying → a single consensus forecast.
- Bias is ignored → the forecast is always above actual → compute bias and correct.
- Finance joins at the end → the decision does not survive cash → the financial roll-up is part of the cycle from the start.
- Too many participants and slides → accountability is lost → a short pack and a fixed membership.
- Forecast accuracy is computed only on the aggregate → errors cancel and SKU purchasing misses → compute WAPE by purchase cells and show both figures.
- The meeting decisions are not logged → a month later nobody remembers who promised what → minutes within the day with an owner and a due date.
Russian specifics
- Russian restrictions on imports, payments and delivery times change fast: revisit S&OP assumptions on FX, duties and logistics every cycle.
- Foreign currency payments are subject to currency control: allow for terms in the cash plan. Rule: Law 173-FZ of 10.12.2003: Federal Law of 10.12.2003 No. 173-FZ on currency regulation and control (rule check not completed).
- S&OP decisions on price reductions on a platform take into account the order of the platform economy law: a price reduction at the partner expense is possible by contract with at least five business days notice and with the consent of the partner (289-FZ, Art. 9, the text was read in the register). Rule: Law 289-FZ of 31.07.2025: Federal Law of 31.07.2025 No. 289-FZ on the platform economy.