1. Home
  2. Blog
  3. How to Forecast Demand in ERP and Run a Monthly S&OP Cycle

How to Forecast Demand in ERP and Run a Monthly S&OP Cycle

  • 7 min read
  • Tranquil team

Demand planning in ERP starts from a simple question the purchasing manager asks every week. How much of each item will we sell before the next shipment arrives? Most distributors answer it from memory and last month’s sales, which works until Ramadan moves, a large customer changes supplier or a product is discontinued. Then the warehouse is full of the wrong items and short of the right ones at the same time.

A forecast does not need advanced software to be useful. It needs clean sales history from the ERP, a method that suits each group of items, a monthly review where sales and supply agree one number, and a way to turn that number into reorder levels and purchase orders.

Independent and dependent demand

Independent demand comes from customers: finished goods, trading items, spare parts sold over the counter. It has to be forecast. Dependent demand is the raw material and components used to make those products, and it is calculated from the production plan and the bill of materials, not forecast separately.

Forecasting raw materials from their own consumption history is a common mistake in small manufacturers. It double-counts the uncertainty, because the finished goods forecast already carries it, and it misses planned changes such as a new product that uses the same material. Spare parts that are both used internally and sold to customers are best split into the two streams.

Classifying items before forecasting

No single method suits every item. Two classifications decide how much effort each item deserves. ABC ranks items by their share of sales value, and XYZ ranks them by how much their monthly demand varies. The cut-off points are the company’s choice. A typical starting point is A for the items that make up most of the sales value, and X for items whose monthly sales vary little around the average.

Class What it means Planning approach
AX High value, steady demand Statistical forecast, reviewed monthly item by item
AZ High value, erratic demand Forecast with sales input, firm customer orders where possible
CX Low value, steady demand Simple reorder level, reviewed quarterly
CZ Low value, erratic demand Order on demand or hold a small safety stock

The classification should be refreshed at least twice a year. New items need their own rule, usually the history of a similar item, until they have enough sales of their own.

Building the statistical forecast

For most trading items, two methods cover the ground. A moving average of the last three to six months suits steady items with no trend. Exponential smoothing gives more weight to recent months and, in its seasonal form, handles both a trend and a repeating yearly pattern. More complex models need several years of clean history and someone who can maintain them.

In Saudi Arabia the seasonal pattern needs one adjustment that standard tools miss. Ramadan, Eid and Hajj follow the Hijri calendar, which is about 11 days shorter than the Gregorian year, so they move earlier every year. A model that reads monthly Gregorian history will put last year’s Ramadan peak in the wrong month. Tag those weeks in the history, or plan the affected items on Hijri periods.

The statistical forecast is then adjusted for what history cannot show, such as a planned promotion, a tender the company expects to win or a competitor leaving the market. Each adjustment is recorded with its reason and owner, so the review can check later whether it helped.

Measuring forecast accuracy: a worked example

The figures below are hypothetical, for three items in one month.

Item Forecast Actual Absolute error Error % of actual
A1 1,000 900 100 11.1%
B7 200 260 60 23.1%
C3 20 10 10 100.0%
Total 1,220 1,170 170

The simple average of the three percentages (MAPE) is 44.7%, which makes the forecast look poor. The weighted measure (WMAPE), total absolute error divided by total actual, is 170 ÷ 1,170 = 14.5%. The difference comes almost entirely from C3, a slow item where an error of 10 units is 100%. That is why MAPE alone misleads on a range with many slow movers.

Bias is the third measure. Total forecast minus total actual is +50 units, or 4.3% above actual. One month means little, but a forecast that is above actual month after month builds stock, and one that is always below causes shortages. Bias is also the quickest way to see whether manual adjustments improve the statistical forecast or just add optimism.

From forecast to reorder levels

The forecast becomes useful when it sets the minimum quantity and the reorder quantity on each item. In a worked example with hypothetical figures, an item sells 40 units a day on average, the supplier’s lead time is 15 days, and the company holds 200 units of safety stock against variation in demand and delivery.

  • Demand during the lead time: 40 × 15 = 600 units.
  • Reorder point (minimum quantity): 600 + 200 = 800 units.
  • Reorder quantity: about one month of demand, 40 × 30 = 1,200 units, rounded to the supplier’s carton or pallet size.

The mistake we see is setting these levels once at go-live and never again. When the forecast for an item rises or falls, its minimum and reorder quantities should follow, or the ERP will keep ordering for last year’s demand. The order-up-to level guide covers the periodic-review alternative.

The monthly S&OP cycle

Sales and operations planning is the monthly meeting where the forecast and the supply plan are agreed. It usually runs in four steps:

  1. Demand review: sales and marketing adjust the statistical forecast and agree one demand number per item group.
  2. Supply review: purchasing, the warehouse and production check whether that demand can be met, and list the gaps such as long lead times, supplier limits or warehouse space.
  3. Reconciliation: planning prepares options for each gap, with their effect on stock, service and cash.
  4. Executive meeting: management chooses between the options and approves the plan for the coming months.

Finance should attend the last step. The approved plan drives purchases, and purchases drive the cash requirement for the next quarter, which is the link to cash flow planning.

What Tranquil does for demand planning in ERP

Tranquil does not replace a forecasting tool. What it provides is the data a forecast is built from and the controls that carry it out. Sales history comes from the Sales reports, including top sales, profit and salesman analysis, and stock by warehouse comes from the Inventory reports. The forecast itself is prepared in a spreadsheet or a planning tool.

The result goes back into Tranquil on each item as a reorder quantity and a minimum quantity. Auto Purchase Order raises purchase orders for the items below their minimum quantity in one step, supplier price lists hold dated supplier prices, and approved sales orders reserve stock, so free stock is visible when salespeople quote. For manufactured items, bills of materials hold the components and quantities. The inventory module is described on the inventory management page, and replenishment rules are covered in inventory restocking.

Frequently asked questions

What is the difference between demand planning and demand forecasting?

Forecasting is the statistical estimate of future sales from history. Demand planning is the wider process that adjusts that estimate with market knowledge, agrees one number across sales and supply, and turns it into reorder levels, purchase orders and production plans.

How much sales history is needed to forecast demand?

For a moving average, a few months is enough. To see a yearly seasonal pattern you need at least two full years, and more for items affected by Ramadan, Eid and Hajj, because those seasons move through the Gregorian calendar.

Which forecast accuracy measure should we use?

Use a weighted error such as WMAPE for the overall picture, because it gives large items more weight, and track bias separately to see whether the forecast is consistently high or low. Simple MAPE is distorted by slow-moving items with small actual quantities.

Does a small distributor need S&OP?

It needs the habit, not the formality. A monthly hour where sales, purchasing and finance look at the same forecast, agree what to buy, and record the reasons for any change covers most of what S&OP is meant to achieve.

See what this article covers in Tranquil

In the demo we show what this article covers on Tranquil’s own screens and answer your team’s questions.

Dammam +966 13 848 8080, +966 13 817 7724, Riyadh +966 11 213 1330
[email protected]
Sunday to Thursday, 8 am to 5 pm.