What Is S&OP? The Sales and Operations Planning Process, Step by Step, and Where It Breaks

Sales and operations planning (S&OP) is a monthly, cross-functional process in which sales, operations, supply chain, and finance agree on one operating plan. Each cycle reconciles the demand plan with the supply plan, the inventory position, and the financial plan. Most companies look 18 to 24 months ahead, with the most detail in the first three to six.

A few distinctions worth making up front:

• A typical cycle runs in five steps: product and portfolio review, demand review, supply review, reconciliation (often called pre-S&OP), and the executive S&OP meeting.

• S&OP is a management process. Software can support it, but buying a tool does not give you one.

• Integrated business planning (IBP) is the same cycle with finance and strategy pulled further in, usually over a longer horizon.

• Demand planning feeds S&OP. It produces the unconstrained forecast that the demand review starts from.

Picture the executive S&OP meeting at a mid-sized consumer goods company. Demand planning brings a forecast. Sales says it’s light because two key accounts are expanding. Operations says it’s heavy because a co-packer just lost a line for three weeks. Finance needs the total to land near the number that went to the board in January. Ninety minutes later the room settles on a figure somewhere in the middle, and each function goes back to its own spreadsheet with its own private view of what will actually happen.

That meeting is the part of S&OP most people see, and it’s only the last step. A systematic review that screened 271 academic papers and analyzed 55 of them in depth describes S&OP as a tactical process that links strategy to execution by pulling together the plans each function would otherwise run on its own (Thomé et al., 2012).

This guide covers how the cycle runs and where it starts to strain in companies whose networks have grown faster than their planning teams.

How is S&OP different from IBP, APS, and MPS?

S&OP and IBP get used interchangeably, and in practice the line between them is blurry. APS and MPS are different things, but they come up in the same conversations, so it helps to know where each one fits.

Where did S&OP come from?

S&OP grew out of manufacturing in the 1980s as a planning layer above MRP II. Richard Ling and Walter Goddard’s book Orchestrating Success (1988) is usually credited with formalizing it. The problem it was designed for explains most of its structure. Sales had a forecast, manufacturing had a capacity plan, finance had a budget, and distribution had its own view of what was moving. Those numbers lived in separate systems that couldn’t reconcile with each other, so people reconciled them once a month in a meeting. The meeting was the integration.

That structure has lasted close to forty years, and for plenty of businesses it still works well. It gets harder to run as SKU counts and locations grow, for reasons covered below.

What are the five steps of the S&OP process?

Most S&OP cycles run in five stages over the two or three weeks after month-end close, so each step works from fresh actuals. Some companies fold the portfolio review into the demand review, and smaller businesses often combine reconciliation with the executive meeting.

The first four steps are preparation. Their job is to hand the executive meeting a short list of decisions with a cost attached to each. When that preparation is thin, the executive meeting ends up re-running the demand and supply reviews, and the decisions slip to next month.

The part of the process that lives in people’s heads

Ask an experienced demand planner what makes the demand review worth attending, and the answer is rarely the statistical baseline. It’s usually something like this:

• The account manager who mentioned that a customer is opening two distribution centers in the spring

• The plant supervisor who knows a line has been running hot and won’t hold through the quarter

• The promotion a regional team ran without logging it anywhere

• The customer’s buyer who has stopped returning calls

None of that is in the ERP. It reaches the plan through conversation, and the reason a number moved is rarely written down anywhere.

There is good evidence that this knowledge improves forecasts. A study of more than 60,000 forecasts at four supply chain companies compared each statistical forecast with the final figure after a planner adjusted it. At three of the four companies, the adjustments improved accuracy on average (Fildes, Goodwin, Lawrence, and Nikolopoulos, 2009). Planners knew things the statistical model didn’t.

The same study found that the value was uneven:

• Large adjustments tended to improve accuracy. Small ones often made it worse, which suggests planners were tinkering when they had no real information to add.

• Upward adjustments helped much less often than downward ones, and more often moved the forecast in the wrong direction. The authors attributed this to optimism bias.

So the knowledge that arrives in the meeting is valuable, and it gets applied inconsistently. Planners can’t be expected to fix that on their own. When the only record of a judgment call is a changed cell in a spreadsheet, nobody can go back a quarter later and see which calls were worth making.

Why does S&OP break down in complex supply chains?

The problems below show up most often in companies with complex networks and small planning teams. None of them means the team is running S&OP badly. They come from the structure of the process.

Consensus averages away the signal. The cycle is built to end in one number. If sales says 1,200 units and operations says 800, the usual outcome is something near 1,000. Both views carried information, though. Sales was high because of a customer conversation, and operations was low because of a supplier problem. The consensus number reflects neither, and by the time anyone reviews the result, the reasoning behind both has usually been lost.

The cadence is monthly and the business is not. A plan agreed in the second week of March gets executed in late April, against conditions nobody in the March meeting could see. A tariff change or a retailer’s shelf reset won’t wait for the next cycle. Most companies respond with off-cycle reviews, which is a sensible workaround and also a sign that the monthly rhythm no longer matches the business.

Everyone arrives with one plan. The reconciliation step is supposed to produce options. In practice it usually produces one recommendation and maybe a fallback, because each extra scenario takes days to build in a spreadsheet and the team doesn’t have those days. So the executive meeting debates the plan in front of it, and the decision can only be as good as the handful of futures the team had time to model.

The plan changes the data it will be measured against. This problem gets less attention than the others, and in our experience it does the most lasting damage. Decisions made in the S&OP cycle get written into the transaction history as though they were market behavior. If you allocate a constrained product away from a customer, the history shows that customer wanting less. If you move an order from a plant at capacity to one with room, the history shows demand shifting between plants. Next cycle’s baseline is fitted to that history, and the error carries forward.

Operations researchers have a name for part of this problem: censored demand. When supply runs short, recorded sales reflect what you shipped, which is less than customers wanted, so a forecast fitted to sales history underestimates true demand unless it corrects for the gap (Nahmias, 1994). Allocation and rerouting make it worse, because they move demand around in the record as well as truncating it. A related effect is well documented between companies. Research on the bullwhip effect showed how order variance grows upstream when each party forecasts from orders that earlier decisions have already shaped (Lee, Padmanabhan, and Whang, 1997). We’ve written about what the internal version looks like at one manufacturer, where a single order rerouted away from a plant at capacity distorted the cycles that followed: Why Predicting Isn’t Understanding.

This is also why forecast accuracy can look healthy while this is going on. The forecast matched the outcome because the plan produced the outcome. Fit a model to that record and you forecast your constrained self, and each cycle inherits the constraints of the one before. Growth gets planned out of the business one cycle at a time.

What does a better S&OP cycle look like?

Each of those problems traces back to one constraint. The reconciliation work is done by people, in the days before the meeting, in tools that can carry one plan at a time. Whatever doesn’t get modeled gets settled by argument in the room.

When the team can model many plans before the meeting, the cycle works differently. Speed matters here only because it lets the team evaluate more options, and more options are what improve the decision. Most of what follows is good practice with any tools.

Start by writing down the knowledge. When a planner learns that a customer is opening two DCs in the spring, that belongs in the plan as a stated assumption with an owner and a date. Everyone downstream can see it, and six months later the team can check whether it held. The Fildes findings point to exactly this kind of record, since it’s the only way to separate the judgment calls that add accuracy from the ones that don’t.

Then answer the executives’ questions before the meeting starts. Most of them are predictable: what happens to cash and service if a promotion lands 30% above plan, or if a tariff takes effect in June instead of September. With those modeled ahead of time, the meeting becomes a choice between costed options. The conversation shifts from whether 1,200 is the right number to a decision with terms attached. For example, the team might carry six weeks of cover on twelve high-risk SKUs, accept 94% service on the long tail, and revisit if the supplier confirms by the 12th.

Finally, keep the scenarios you turned down. At quarter end, compare what happened with the plan you chose and with the plans you rejected. Over a few quarters that comparison shows whether the team leans cautious or optimistic, which no accuracy metric will tell you.

For some businesses, this can wait. A company with a few hundred SKUs, one or two plants, and steady demand can run a perfectly good S&OP cycle in spreadsheets, and its effort is better spent on meeting discipline than on new technology.

None of this requires giving up S&OP. The monthly cadence and the executive forum still do their jobs. What moves is the modeling, which happens ahead of the meeting so the ninety minutes can go to deciding.

What should you look for in S&OP technology?

Whatever you evaluate, four capabilities matter more than most feature lists suggest.

The model should reflect your network. Most APS products ship a fixed library of model shapes, the same for every customer, and fit them to your history. Your plants, lanes, substitution rules, and capacity constraints are represented only loosely. The achievable accuracy was effectively fixed before the project started, and that puts a ceiling on what any scenario can tell you. The forecast can also be aimed at the wrong target before modeling begins, which we covered across make-to-stock, make-to-order, and engineer-to-order networks in Every Supply Chain Needs a Different Forecast.

Scenarios need to be cheap. If each scenario takes a day to build, the team will build two, and the executive meeting will debate one of them. The faster scenarios run, the more of the executive team’s questions get answered before the meeting.

It has to separate market signals from the effects of your own decisions. Otherwise a rerouted order and a lost customer look the same in the history, and the next cycle inherits the error.

Planner knowledge needs a place to live. The customer expansion and the supplier problem should enter the plan as inputs the model uses, with the reasoning attached, so the team can see later which ones paid off.

This is the problem Omnifold works on. Omnifold builds reasoning AI for supply chain planning. You get your own planning platform, containing your model: a reasoning model of your supply chain, trained with reinforcement learning to reduce forecast error and directed at an economic objective. The model works from three inputs we call the Context Gap: history (what happened), structure (how your network works), and knowledge (what your people know is changing). Planners add knowledge as Enhancements, the platform proposes Enhancements of its own, and the team can question the plan directly in Conversations.

The payoff shows up most in the parts of S&OP that are hardest to get right, the portfolio and demand reviews, where new products and new markets enter the plan with little or no history. Eric Falkenmayer, Senior Manager of Demand Planning at Not Your Mother’s, sees forecasting for new SKU launches, new customer expansions, and new channels as the highest-value application of AI in planning, ahead of incremental accuracy gains on the base business. When the brand launched in a new country with no local sales history, Omnifold’s forecast for the launch and the following six months was about 80% accurate.

How do you know if your S&OP process is working?

Put these questions to your planning team. The answers usually point to the specific problem faster than a maturity assessment will.

1. How many scenarios were modeled before the last executive S&OP meeting? If the answer is one, the meeting was a negotiation.

2. When a planner overrides the statistical forecast, is the reason recorded anywhere a colleague could read it next quarter?

3. Can you say what the plan you rejected last quarter would have produced?

4. How many decisions made in the cycle get written back into the demand history as though they were market behavior?

5. How long does it take to answer a question the executive team asks in the meeting? If the usual answer is next month, the cadence is setting the pace of the business.

6. Does the cycle end with a number, or with a decision that has a cost and a review date attached?

Frequently asked questions

What is the difference between S&OP and IBP? IBP extends S&OP to include finance, strategy, and portfolio decisions over a longer horizon. The process mechanics are largely the same. The useful question is whether finance takes part in the cycle or only receives its output.

What is the difference between S&OP and demand planning? Demand planning produces the unconstrained demand forecast, which is the input to the demand review in step two of the S&OP cycle. S&OP takes that forecast and reconciles it with the supply, inventory, and financial plans. Demand planning is one function’s job. S&OP is the cross-functional process that decides what the business will do about the forecast.

What metrics should you track for S&OP? Most teams track forecast accuracy and bias at the level where decisions get made (SKU and location, as well as product family), service level or fill rate, inventory days or turns, and plan adherence, meaning how closely execution followed the agreed plan. It also helps to measure the process itself, such as how many decisions the executive meeting made versus deferred. Read forecast accuracy alongside service and lost sales, because accuracy can look healthy while the plan is constraining demand.

What is an S&OP maturity model? A maturity model scores an S&OP process in stages, typically from informal, siloed planning up to fully integrated planning tied to strategy and finance. There’s no single standard; the 2012 research synthesis noted a lack of unifying frameworks for S&OP maturity models (Thomé et al., 2012). A maturity model is useful for benchmarking. The checklist above is a faster way to find the specific constraint holding your process back.

Should S&OP be monthly or weekly? Monthly is the convention because the cycle was built around a month-end close. The right cadence depends on how fast your demand and supply conditions change and how long it takes you to produce a plan. Many companies keep the monthly executive meeting and run a weekly demand and supply check underneath it.

Who should own S&OP? The cycle needs a process owner who does not own any single plan, commonly an S&OP or IBP manager sitting in supply chain but reporting outside a single function. The executive meeting should be chaired by whoever can commit the business, usually the general manager or president.

How long does it take to stand up an S&OP process? A basic cycle can be running in one to two quarters. Getting to the point where decisions are made in the executive meeting, and not escalated out of it, usually takes longer, because it depends on trust in the numbers.

Does S&OP improve performance? The evidence is thinner than the category’s confidence suggests. The 2012 research synthesis found that the most common expected outcome was cross-functional integration of plans. Of the 55 papers it analyzed, just six measured the impact of S&OP on the performance of the firm (Thomé et al., 2012). In our experience, the companies that get the most from S&OP use the executive meeting to make and record decisions, and spend little of it reviewing last month.

Related reading

• Why Predicting Isn’t Understanding: what happens when a forecast can’t tell your own decisions apart from market behavior.

• Every Supply Chain Needs a Different Forecast. Your APS Gives Everyone the Same One.: why the right forecast target differs across make-to-stock, make-to-order, and engineer-to-order networks.

• What Not Your Mother’s Haircare Is Prioritizing in 2026: why a fast-growing CPG brand sees innovation forecasting as the highest-value use of AI in planning.

Sources

• Thomé, A. M. T., Scavarda, L. F., Fernandez, N. S., and Scavarda, A. J. (2012). Sales and operations planning: A research synthesis. International Journal of Production Economics, 138(1), 1–13.

• Fildes, R., Goodwin, P., Lawrence, M., and Nikolopoulos, K. (2009). Effective forecasting and judgmental adjustments: an empirical evaluation and strategies for improvement in supply-chain planning. International Journal of Forecasting, 25(1), 3–23.

• Lee, H. L., Padmanabhan, V., and Whang, S. (1997). Information Distortion in a Supply Chain: The Bullwhip Effect. Management Science, 43(4), 546–558.

• Ling, R. C., and Goddard, W. E. (1988). Orchestrating Success: Improve Control of the Business with Sales & Operations Planning. Oliver Wight Limited Publications.

• Nahmias, S. (1994). Demand estimation in lost sales inventory systems. Naval Research Logistics, 41(6), 739–757.