Sales is confidently forecasting a big quarter. Operations is building to a different number it trusts more. Finance is reporting a third figure to the board. None of them is lying, they are each planning from the data they happen to own, and the three plans will collide somewhere expensive: a stock-out, a warehouse full of the wrong thing, or a revenue miss nobody saw coming. Sales and operations planning (S&OP) is the unglamorous monthly ritual that drags those competing numbers into one room and refuses to let people leave with different ones.

The quick version

  • S&OP is a recurring (usually monthly) process where sales, operations and finance agree one set of numbers for demand and supply over the next several months, replacing the private forecasts each function used to run on.
  • Demand planning is the front half: a best estimate of what customers will actually buy, blending history, market signals and judgement, not just last year plus a hopeful percentage.
  • The classic shape is a five-step cycle: gather data, plan demand, plan supply, reconcile the gaps, then sign off in an executive meeting.
  • The point isn't a perfect forecast, it's a shared one. A single number everyone argued over beats five accurate-looking numbers nobody reconciled.

The idea in depth: one set of numbers, argued over monthly

S&OP has a surprisingly precise origin. The term and its modern meaning were coined in the 1980s and are generally attributed to Richard "Dick" Ling, a consultant with the firm Oliver Wight, who set it out with Walter Goddard in their 1988 book Orchestrating Success. The problem Ling was naming is older than the term: a company's commercial side and its operational side run on different clocks and different incentives, so they end up planning from different assumptions. S&OP is the mechanism that reconnects them, described by its practitioners as the process through which an executive team "continually achieves focus, alignment and synchronization" across functions.

The mechanism is a monthly cycle, usually drawn as five steps: gather the data, run a demand review, run a supply review, reconcile the gap between the two, and take the result to an executive meeting for sign-off (the canonical five steps are data gathering, demand planning, supply planning, a pre-meeting and an executive meeting). It is a tactical process, the typical horizon runs out to around 18 months at the product-family level, not the next sprint and not the five-year strategy.

flowchart LR
  A(["1 · Gather data
sales, stock, actuals"]) --> B(["2 · Demand review
what will customers buy?"])
  B --> C(["3 · Supply review
what can we make/serve?"])
  C --> D(["4 · Reconcile
close the demand–supply gap"])
  D --> E(["5 · Executive sign-off
one agreed plan"])
  E -.->|"next month"| A
The classic monthly S&OP cycle. The dotted return is the discipline: it runs every month, not once a year. Leaders Loop

The shift is to stop treating "the forecast" as a number one team hands to another, and to start treating it as a decision the leadership team makes together, on a fixed cadence. If your sales, operations and finance leads cannot point at a single agreed demand number for next quarter, you don't have an S&OP process. You have three forecasts and a coming argument. The first practical step is a humble one: book a standing monthly meeting where those three people reconcile their numbers in front of each other, and nobody is allowed to walk out still privately planning to a different one.

Demand planning: the forecast is a conversation, not a calculation

The front half of S&OP is demand planning, the estimate of what customers will actually buy. This is where most of the value, and most of the self-deception, lives. The tempting approach is to extrapolate: take last year's sales, add a growth percentage, call it a plan. The discipline of demand planning is to treat that statistical baseline as a starting point that human judgement then adjusts, a known promotion, a lost customer, a competitor's launch, a price change none of which the history can see.

There is a useful, counter-intuitive piece of evidence to keep you honest here. Across four decades of the Makridakis (M) forecasting competitions, open contests run by Spyros Makridakis from 1982 onward to compare forecasting methods on thousands of real series, a consistent finding emerged: statistically sophisticated methods do not necessarily beat simple ones, and combining several methods tends to beat any single one. (Machine-learning methods only decisively overtook classical statistics in the M5 competition, which ended in 2020.) The lesson for a demand planner is liberating: you don't need an exotic model to start. A simple baseline, openly adjusted by the people who know the market, and sense-checked against a couple of other methods, is a perfectly respectable forecast.

A single number everyone argued over beats five accurate-looking numbers nobody reconciled.

What follows from this is simple to say and harder to do: make the forecast a named, owned, written-down judgement, and then measure how wrong it was. Pick a simple accuracy metric, review last month's forecast against what actually sold, and ask why the gap happened before you set the next one. Forecasting accuracy improves not by buying a cleverer algorithm but by closing this loop: forecast, compare, learn, adjust. A forecast nobody checks afterwards is a wish.

Why the gaps amplify: the bullwhip effect

There is a structural reason demand planning is hard, and it has a name. In their 1997 MIT Sloan Management Review article "The Bullwhip Effect in Supply Chains," Hau Lee, V. Padmanabhan and Seungjin Whang showed that small wobbles in real customer demand get amplified as they travel up a supply chain, a retailer's modest order swing becomes a distributor's bigger one and a factory's wild one. Their famous case was Procter & Gamble's Pampers: babies are a remarkably steady source of demand, yet orders upstream lurched all over the place.

Crucially, they found this isn't caused by stupidity. Four rational behaviours drive it: each link forecasting from the orders of the link below it rather than from real end demand; batching orders to save on ordering costs; forward-buying when prices are promoted; and inflating orders during shortages to game an allocation. Each decision is individually sensible; together they distort the signal. The defence is to plan from demand that sits as close to the real customer as you can get, actual point-of-sale or usage data, not the orders your own downstream has already filtered. The deeper reason S&OP insists on one shared number is exactly this: every place the signal is re-interpreted privately is a place the bullwhip cracks.

An honest limitation. S&OP is not a forecasting machine and won't make an uncertain future certain. Its weak spot is human, not technical: the executive meeting can decay into a status update where nobody actually decides anything, or the agreed "plan" can be quietly overridden by whoever shouts loudest. The same firm, Oliver Wight, later coined Integrated Business Planning (IBP) in the early 2000s partly to fix that, pulling finance and strategy fully into the room and stretching the horizon out past two years. Whether you call it S&OP or IBP matters far less than whether the meeting produces a decision people are held to. A cadence without consequences is theatre.

A worked example

Take a mid-sized maker of premium coffee equipment, call it Cromwell & Vale. (Illustrative figures throughout; this is a teaching example, not real data.) Sales is forecasting a strong run into the December gifting season. Operations, burned by last year's leftover stock, is building cautiously. Finance has pencilled a third number into the board pack. Nobody is wrong on their own terms; nobody is planning from the same sheet.

flowchart TD
  S(["Sales says:
+30% for December"]) --> R{"S&OP reconciliation
one agreed number?"}
  O(["Operations says:
build flat, avoid leftover stock"]) --> R
  F(["Finance says:
+12% in the board pack"]) --> R
  R --> P(["Agreed plan: +18%,
with a flex option held"])
  P --> G(["Demand review finds the
real driver: 2 wholesale accounts"])
  G --> H(["~£200k stock-out risk avoided
(illustrative), no panic buy"])
Three private numbers reconciled into one agreed plan, and then pressure-tested for what's actually driving the demand. Leaders Loop

They run the cycle. In the demand review, instead of arguing about the headline percentage, they ask what is driving the optimism, and find that most of Sales' +30% rests on two large wholesale accounts that haven't actually confirmed. Strip the unconfirmed orders and the defensible baseline is nearer +18%. In the supply review, operations confirms it can build to +18% but not +30% without paying for air-freight and overtime. In reconciliation, the gap is named honestly: the upside depends on two specific deals, so they agree a base plan of +18% and hold a pre-costed "flex" option to ramp only if those accounts sign by a set date. The executive meeting signs off that single plan, one number, one trigger, one owner.

The payoff isn't a clairvoyant forecast. It is that, in this illustrative case, the firm avoids both failure modes: it doesn't over-build to a fantasy +30% (and eat the leftover stock that scared operations), and it doesn't under-build to a timid number and miss a real season. The bullwhip is dampened because everyone is now planning from the same end-demand picture, and the board pack finally matches the operational plan. Same information, one version of it.

Frequently asked questions

What's the difference between S&OP and demand planning?

Demand planning is one input to S&OP, specifically the best estimate of what customers will buy. S&OP is the wider process that takes that demand plan, sets it against what the business can actually supply, reconciles the gap, and gets leadership to commit to a single plan. You can do demand planning in isolation, but without the supply and reconciliation steps it stays a forecast, not a decision.

How is this different from budgeting or the annual plan?

The annual budget is a once-a-year financial target, often political and quickly out of date. S&OP is a rolling tactical process, usually monthly, typically looking out around 18 months, that keeps adjusting the operational and demand plan as reality moves. The budget says what you hope to earn; S&OP keeps the operational machine pointed at what's actually happening. Good practice connects the two so the rolling plan and the financial picture don't drift apart.

What's the difference between S&OP and IBP?

Integrated Business Planning (IBP) is essentially mature S&OP with the scope widened. Oliver Wight, who coined both terms, frames S&OP as the operational foundation and IBP as the evolution that brings finance and strategy formally into the room and stretches the horizon out past two years. For most teams the distinction matters less than the discipline: one cadence, one agreed plan, real decisions. Get that working and what you call it is a footnote.

Do I need expensive forecasting software to start?

No. The Makridakis competitions found that simple forecasting methods often match sophisticated ones, so the algorithm is rarely your bottleneck, the missing meeting is. You can run a credible first S&OP cycle on a spreadsheet and a standing monthly slot. Buy better tooling once the process exists and you've hit the limits of what a spreadsheet can hold, not before.

We're a service business, not a factory, does this apply?

Yes. "Supply" becomes capacity, consultants, support agents, engineers, clinic hours, and "demand" becomes the pipeline of work coming in. The reconciliation is the same: are we about to promise more than we can staff, or hire for work that isn't coming? Any business that has to match incoming demand against finite capacity benefits from a shared monthly planning rhythm.

Related in the Toolkit

S&OP doesn't stand alone. The demand-and-supply balancing it does is the planning layer that feeds your supply chain and sourcing decisions, and the agreed plan only means something if it shows up in the numbers, which is where the financial statements come in.

Where to go next