Sales and demand forecasting for small retailers: a practical guide

Every reorder is a small bet on the future. Order too little and customers find an empty shelf. Order too much and cash sits in the stockroom while the product nears its expiry date. A sales forecast does not remove the bet, but it replaces a guess with an estimate you can check.

This guide explains what a forecast is, how to turn one into practical reorder decisions, and how to avoid the most common mistakes. It works whether you forecast in a notebook, a spreadsheet or a retail system.

What a sales forecast is

A forecast is an estimate of how much you are likely to sell over a coming period, based on what you have sold before. Most retail forecasts start from the sales rate: the average quantity sold per day over a recent period.

From there, a forecast can take account of patterns you already know:

• day of the week, if Saturdays are always busier;
• time of the month, if sales rise around paydays;
• seasons and events, such as school terms or public holidays;
• trend, if a product is steadily selling more or less than before.

A forecast is always a likely range, not a promise. Its job is to help you prepare, not to predict the future exactly.

Choose the right horizon

Different decisions need different horizons:

• 7 days: staff rotas, fast-moving goods and products delivered quickly.
• 14 days: most regular reorders, especially when suppliers deliver weekly.
• 30 days: slower movers, imported goods with long lead times, and cash planning.

The further ahead you look, the wider the uncertainty. Revisit longer forecasts as new sales come in.

Forecast products and categories

Forecast the products that matter most first: the fast movers, the high-value items and the products customers come in specifically to buy. For the long tail of slow movers, a forecast by category ("cooking oils", "baby care", "phone accessories") is often more reliable than one per product, because there is more data behind it.

If you run several branches, forecast each branch separately. A product that flies off the shelf in one neighbourhood may barely move in another.

From forecast to reorder point

A forecast becomes useful when it tells you when to reorder. The classic method uses three numbers:

• average daily sales of the product;
• lead time: the number of days between placing an order and having the goods on the shelf;
• safety stock: a buffer for days when sales are higher or deliveries are later than usual.

The reorder point is the stock level at which you should place a new order:

Reorder point = average daily sales × lead time + safety stock

When stock on hand, plus anything already on order, falls to the reorder point, it is time to reorder.

Illustrative example (the numbers are made up to show the method, not real figures). You sell an average of 12 bags of rice a day. Your supplier delivers 4 days after you order. You keep 2 days of sales as safety stock, which is 24 bags. The reorder point is 12 × 4 + 24 = 72 bags. You have 80 bags on the shelf and none on order, so you do not need to order today. When stock reaches about 72 bags, place the order.

Projected stockout date

The same numbers tell you when a product will run out if you do nothing. Divide the stock on hand by the average daily sales. In the example, 80 bags ÷ 12 a day is just under 7 days. If the supplier needs 4 days, you have about 2 days in which to place the order.

Setting safety stock

There is no perfect safety stock. A simple, practical approach:

• keep more for products customers cannot do without, and for suppliers who are often late;
• keep less for products that expire quickly or are expensive to hold;
• review it when you see repeated stockouts (too little) or stock that sits for weeks (too much).

Use the supplier, not just the shelf

A reorder decision also depends on the supplier. Track for each supplier how often deliveries arrive late, short or with substitutions. A supplier who is regularly two days late needs a longer lead time in your calculation, or a higher safety stock. If one supplier often lets you down on an important product, it is worth knowing which alternative suppliers carry it.

Common forecasting mistakes

• Counting stockouts as low demand. If a product was out of stock for three days, it sold nothing, but demand was not zero. Leave those days out, or the forecast will be too low.
• Ignoring promotions. A promotion lifts sales temporarily. Do not treat that week as the new normal.
• Trusting a forecast for a new product. With little history, a forecast is weak. Start cautiously and review often.
• Forgetting open orders. Stock already ordered but not yet received should be counted before you reorder again.
• Setting and forgetting. Compare forecasts with actual sales every week or two and adjust.

A weekly forecasting routine

1. Review the fast movers and high-value products whose stock is close to the reorder point.
2. Check the projected stockout dates for anything due to run out before the next delivery.
3. Look at open orders and supplier reliability before placing new orders.
4. Draft the orders, review them with the person responsible for purchasing, then send them.
5. Note anything unusual (a promotion, a holiday, a delivery problem) so you can read next week's numbers correctly.

Frequently asked questions

How much sales history do I need?

For a product that sells every day, a few weeks gives a usable sales rate. For seasonal patterns, you need at least one full season. Slow movers need longer, which is why forecasting them by category often works better.

Should a system place orders for me automatically?

We recommend not. A system can calculate reorder points and draft purchase orders, but someone should review them before they are sent: they know about the delayed truck, the price change or the promotion next week.

What if my forecasts are often wrong?

Check the records first: unrecorded sales, unchecked deliveries and duplicate products all distort forecasts. Then check for stockout days and promotions in the history. Accurate records improve forecasts more than any formula.

Conclusion

Forecasting does not need to be complicated. A sales rate, a realistic lead time and a sensible safety stock already give you reorder points and stockout dates you can act on. Review them weekly, compare them with what actually happened, and keep the final decision with your team.

In NexaPOS, ML Insights provides 7-, 14- and 30-day sales forecasts, demand by product and category, projected stockout dates, safety-stock guidance and reorder recommendations. Purchase orders are created as drafts for your team to review; they are never submitted automatically.