What Is Demand Planning?
Demand planning is how a company estimates what customers will order of each item over the coming period, and then turns that estimate into decisions: how much to buy, how much to produce, how much to hold in each warehouse, and when. It is the bridge between what sales expects and what purchasing, production and the warehouse actually do.
Demand planning and demand forecasting are often used interchangeably, but they are not the same. Forecasting is one part of planning: the statistical number that answers "how much will we sell, based on what we sold before?" Planning takes that number and adjusts it with what the team knows and history does not: an upcoming promotion, a large new customer, a competitor raising prices, or Ramadan arriving about eleven days earlier on the Gregorian calendar than last year. Then it converts the adjusted number into a purchasing, production and inventory plan everyone works to.
This guide covers the stages of demand planning, the core forecasting methods with worked numbers, how to measure forecast accuracy, how to handle seasonality in the Saudi market, and how to run a monthly cycle that ties sales to operations and finance.
Why Demand Planning Matters
Every purchase or production decision is a demand forecast, written down or not. A buyer ordering a thousand cartons expects them to sell. The question is whether that expectation is structured, data-based and shared across departments, or sitting in one person's head.
- Fewer stockouts: knowing demand will rise lets you order on time instead of paying for rush freight after the shelf empties.
- Less excess stock: buying to a considered forecast rather than to fear frees cash that would otherwise sit in the warehouse.
- Better supplier terms: a supplier who sees your needs months ahead can offer better prices and priority than one hit with urgent orders.
- Capacity planning: the plant knows when it needs an extra shift; the warehouse knows when it needs temporary labor.
- Cash planning: finance knows when large purchase payments will fall due and can arrange funding or negotiate terms.
For what poor planning does on both sides, see our article on overstocking and understocking.
The Stages of Demand Planning
1. Collect and clean the data
The starting point is sales history by item, branch, customer and month. Raw data needs cleaning first:
- Sales are not demand: if an item was out of stock for three weeks, sales in that period were zero but demand was not. Record stockouts and turned-away orders, or your model will learn that demand is low.
- One-off events: a single government tender or a heavy clearance sale inflates one month. Flag it and exclude it from the baseline or treat it separately.
- Returns: use net sales after returns, especially in sectors with high return rates.
- Replacement items: when a new SKU replaces an old one, link their histories.
2. Build a statistical forecast
Once data is clean, calculate a baseline forecast with a method that suits each item's pattern. The next section walks through the main methods.
3. Enrich with market knowledge
Sales and marketing add what they know: planned promotions, customers in negotiation, launches, price changes. Each adjustment is recorded with its reason so you can later check whether it improved the forecast.
4. Agree on one plan
Sales tends to be optimistic, finance conservative, purchasing cautious. A monthly meeting that gets everyone to one number is the heart of the process.
5. Execute and monitor
The plan becomes purchase orders, production orders and target stock levels. Each month actual sales are compared with the forecast, error is measured, and methods are adjusted where needed.
Core Forecasting Methods With Worked Numbers
We will use one hypothetical item throughout: bottled water sold by a distributor, with the last six months of sales in cartons at 1,000, 1,100, 1,050, 1,200, 1,250 and 1,300.
Simple moving average
Average the last few months, say three: (1,200 + 1,250 + 1,300) ÷ 3 = 1,250 cartons for next month.
Easy to understand, but it lags the trend. Sales are clearly rising and the average still sits below last month.
Weighted moving average
Give recent months more weight. With weights of 50%, 30% and 20% from newest to oldest: 1,300 × 0.5 + 1,250 × 0.3 + 1,200 × 0.2 = 650 + 375 + 240 = 1,265 cartons.
Closer to the trend, though choosing weights takes some trial and error.
Exponential smoothing
New forecast = previous forecast + alpha × (actual − previous forecast). If last month's forecast was 1,250, actual was 1,300 and alpha is 0.3: 1,250 + 0.3 × 50 = 1,265 cartons.
A higher alpha reacts faster to change; a lower one is steadier. Extended versions of this method also handle trend and seasonality.
Seasonal indices
If summer sales usually run higher, an average is not enough. Calculate a seasonal index per period: that period's sales divided by the average period. If the monthly average is 1,200 and a typical July is 1,560, July's index is 1.3. With a non-seasonal baseline of 1,265, the July forecast becomes 1,265 × 1.3 = about 1,645 cartons.
| Method | Best for | Weakness |
|---|---|---|
| Simple moving average | Stable items with no clear trend | Lags trends and seasons |
| Weighted moving average | Items changing gradually | Weights are judgment calls |
| Exponential smoothing | Large item counts needing automated calculation | Alpha needs tuning |
| Seasonal indices | Items with recurring seasons | Needs at least two years of data |
| Causal models | Demand driven by price, campaigns or other factors | Needs data and analytical skill |
Newer systems increasingly use machine learning models that weigh many factors at once, which we cover in our article on AI demand forecasting. For most companies, though, the right start is simple methods applied consistently to clean data.
Measuring Forecast Accuracy
Two measures cover most needs.
Mean absolute percentage error (MAPE)
For each period, take the absolute error as a percentage of actual, then average. A hypothetical three months:
| Month | Forecast | Actual | Absolute error | Error % |
|---|---|---|---|---|
| 1 | 1,200 | 1,100 | 100 | 9.1% |
| 2 | 1,150 | 1,250 | 100 | 8.0% |
| 3 | 1,300 | 1,200 | 100 | 8.3% |
MAPE is roughly 8.5%, so accuracy is around 91.5%.
Bias
Sum the errors with their signs. If you over-forecast month after month, you have an optimistic bias that becomes excess stock. If you under-forecast consistently, you get recurring stockouts. Bias is more dangerous than random error because it accumulates in one direction.
Track both measures by item class, starting with A items that carry most of the value. Do not expect the same accuracy everywhere: fast, stable movers forecast far better than slow, intermittent ones.
Seasonality in the Saudi Market
One of the biggest causes of forecast failure in the Kingdom is applying Gregorian monthly indices to demand that follows the Hijri calendar. Ramadan and both Eids move about eleven days earlier each Gregorian year. An index built for March in a year when Ramadan fell in March will mislead you in a year when it falls in February.
Seasons worth modeling separately:
- Ramadan and Eid al-Fitr: a sharp rise in food and beverages before and early in Ramadan, then clothing, gifts and sweets before Eid.
- Eid al-Adha and Hajj: heavy demand in Makkah, Madinah and surrounding areas for food, services and transport.
- Summer: air conditioners and spare parts, water and cold drinks, with dips in some sectors as families travel.
- Back to school: stationery, uniforms, bags and devices.
- Major sale events: White Friday and year-end sales, especially in retail and e-commerce.
- The monthly salary cycle: consumer retail spending typically rises after payday and then eases.
Handling Hijri seasons
- Flag Ramadan and Eid days in your historical sales for each year.
- Measure the seasonal effect relative to those days, not to the Gregorian month, for example the uplift in the two weeks before Ramadan versus normal weeks.
- For each new year, apply that effect to the Gregorian dates the season actually falls on.
- Work back from lead time: if the season starts on a given date and import lead time is sixty days, the purchase decision has to happen at least two months earlier.
The Monthly Planning Cycle
Demand planning works when it becomes a monthly routine with clear owners, often called sales and operations planning (S&OP):
| Week | Step | Owner | Output |
|---|---|---|---|
| 1 | Refresh data and statistical forecast | Demand planner | Baseline forecast per item or group |
| 2 | Demand review with sales and marketing | Sales manager | Forecast adjusted for promotions and new customers |
| 3 | Supply and capacity review | Purchasing, production, warehouse | Purchase and production plan, list of constraints and gaps |
| 4 | Executive meeting | Senior management and finance | Final plan, budget and decisions on gaps |
The most important output is a single agreed number. If sales works to one figure, purchasing to another and finance to a third, you do not really have demand planning, however advanced the statistics.
Not Every Item Is Planned the Same Way
- Fast, stable movers: sold daily in similar quantities. Automated methods like exponential smoothing work, with a quick monthly review. The system can propose purchase requests and the buyer just approves.
- Seasonal items: need seasonal indices and early buying, with a dedicated review before each season that starts earlier than the lead time. The hardest call is the size of the last order before the season ends, because leftovers may wait a full year.
- Slow, intermittent items: spare parts ordered every couple of months in varying amounts. Averages mislead because most months are zero. Min-max levels based on target service level and stockout cost, or buying to order where lead time allows, usually work better.
- New items: no history. Use a similar item as a reference or sales team estimates, order small initial quantities, and review weekly until data builds up.
- Project-driven items: in contracting and project supply, demand comes from the bill of quantities and project schedule more than from history.
From Forecast to Stock Levels
A forecast always carries error. Safety stock protects against that error and against lead-time variability. The more accurate your forecasts, the less safety stock you need, which is one of the biggest financial payoffs of better planning.
Hypothetical example: forecast daily demand is 50 units, lead time is 14 days, and you hold six days of safety stock because demand is volatile. Reorder point = 50 × 14 + 50 × 6 = 700 + 300 = 1,000 units. If better forecasting means four days of safety stock is enough, the reorder point drops to 900 and you permanently free 100 units of stock. Multiply that across hundreds of items and the cash impact becomes clear.
See our guide to safety stock for the calculation options.
Common Demand Planning Mistakes
- Forecasting sales instead of demand: ignoring stockout periods lowers the forecast and causes more stockouts in a loop.
- One forecast for the whole company: demand in Riyadh differs from Jazan, and retail differs from wholesale. Plan at the level where stock decisions are made.
- Undocumented overrides: when a manager adds 20% "on instinct" with no recorded reason, nobody can tell whether overrides help or hurt.
- Targets instead of forecasts: sales targets are ambitious by design. Buying to target means excess stock every time the target is missed.
- Not measuring accuracy: without measurement, you cannot tell which method works for which item.
- Ignoring the product lifecycle: launches have no history and end-of-life items drop quickly; both need special handling.
How an ERP Supports Demand Planning
Spreadsheet planning works for a few dozen items. With hundreds of items and several warehouses, simply collecting and refreshing data eats the planner's time. An ERP helps in three ways:
- Unified, current data: sales, returns, balances and open purchase orders in one place.
- Plan linked to execution: the approved forecast becomes purchase requests or production orders without rekeying.
- Continuous comparison: actuals are compared with the plan automatically, so deviations surface early.
In Tranquil, inventory management, purchase management and sales share demand data and balances across central inventory and branches, and demand reports and purchase approval workflows can be customized to how you work. For manufacturers, material planning helps translate the demand plan into raw material requirements.
Frequently Asked Questions
What is the difference between demand planning and demand forecasting?
Forecasting produces the statistical number from historical data. Demand planning adjusts that number with market knowledge, promotions and new customers, then turns it into an agreed purchasing, production and inventory plan.
How much history do I need?
Simple methods work with a few months, but seasonality needs at least two years and ideally three to separate seasonal patterns from noise.
How should forecasts handle Ramadan?
Do not rely on the Gregorian month. Flag Ramadan and Eid days in your history, measure the effect relative to those days, apply it to the new year's dates, and work back from lead time.
What is a good forecast accuracy?
There is no single target for all items. Stable fast movers achieve much higher accuracy than slow, intermittent ones. Measure where you are today, then work on improving accuracy and removing bias.
Does Tranquil help with demand planning?
Yes. Tranquil brings sales, inventory and purchasing into one system and supports customized reports and approval workflows. Contact the Tranquil team to discuss your planning needs.