Inventory forecasting means estimating how much of each item you'll sell before your next delivery arrives, so you can order enough to avoid stockouts without tying up cash in extra stock. For most small businesses, a practical method is recent average sales adjusted for season, multiplied across supplier lead time, plus a modest safety buffer, then checked against actual sales every week.
This guide is for retailers, e-commerce sellers, cafés, parts counters and anyone who reorders physical goods and keeps swinging between "sold out" and "why is the back room full?" You'll learn a simple forecasting method, how to set reorder points and order quantities, a worked example with real numbers, and how to measure whether your forecasts are improving.
Why ordering by gut feel gets expensive
Running out costs you sales, and sometimes the customer. Overbuying costs you cash, shelf space and often markdowns. Most owners who order by feel make both mistakes at once: too much of slow items they're attached to and too little of fast items they take for granted.
Forecasting doesn't need to be perfect to help. Even a rough, consistent method beats instinct, because you can see where it was wrong and adjust.
The core formulas, in plain English
- Forecast demand per week = recent average weekly sales × seasonal adjustment
- Lead time demand = forecast weekly demand × weeks from placing an order to having it on the shelf
- Safety stock = extra units to cover sales spikes or late deliveries
- Reorder point = lead time demand + safety stock. When stock on hand plus stock on order falls to this level, reorder.
- Order-up-to level = forecast demand × (lead time + weeks between orders) + safety stock. Order enough to bring you back up to this level.
The seasonal adjustment compares a coming period with your average. If you sold 30% more in May than in a typical month last year, May's adjustment is 1.3.
Worked example: a hypothetical bike shop
Suppose a hypothetical bike shop is forecasting inner tubes, one of its fastest movers, heading into spring.
Step 1: Recent average
Over the last 12 weeks the shop sold 480 tubes, an average of 480 ÷ 12 = 40 a week.
Step 2: Seasonal adjustment
Last year, sales in the coming month ran about 30% above the average of the prior 12 weeks. Seasonal adjustment: 1.3.
- Forecast: 40 × 1.3 = 52 tubes a week
Step 3: Lead time and safety stock
The supplier usually delivers in 2 weeks. The shop orders weekly. It decides to hold about one week of forecast demand as safety stock, because deliveries sometimes slip and a sunny weekend can spike sales.
- Lead time demand: 52 × 2 = 104
- Safety stock: 52
- Reorder point: 104 + 52 = 156
- Order-up-to level: 52 × (2 + 1) + 52 = 208
Step 4: Place the order
Today the shop has 110 tubes on hand and 20 already on order, for a total of 130. That's below the reorder point of 156, so it orders.
- Order quantity: 208 − 130 = 78 tubes
- Tubes come in cases of 25, so the shop rounds up to 100 (4 cases).
Step 5: Compare with the old habit
The owner used to order "a few cases" whenever the bin looked low. Some weeks that meant running out on a Saturday. Other times it meant 300 tubes sitting in the back through winter. With a reorder point, the decision is mechanical, and the owner's attention goes to exceptions.
The same logic for a slow, expensive item
Now take a mid-range bike that costs the shop $600 and sells about 10 a month in season. If the owner buys 30 at once for a small volume discount, that ties up $18,000. Buying 10 a month ties up $6,000. The extra $12,000 sits on the floor for weeks, and if the model is replaced next season, any leftovers may need a markdown. Unless the discount clearly beats that risk and cash cost, smaller, more frequent orders win.
Focus your effort with ABC classification
You don't need to forecast every item with the same care. Rank items by annual sales value, meaning units sold × cost, and group them:
| Class | Typically | How to manage |
|---|---|---|
| A | A small number of items that make up most of your sales value | Forecast individually, review weekly, keep tight reorder points |
| B | Middle group | Review every two weeks, use simple averages |
| C | Many low-value items | Review monthly, use generous min/max levels, and don't overthink them |
This keeps forecasting manageable. You might carefully manage 40 A-items and let simple rules handle 800 C-items.
How to measure and improve your forecasts
- Record the forecast and the actual for each A-item every week.
- Calculate the error: |actual − forecast| ÷ actual. If you forecast 52 and sold 60, the error is 8 ÷ 60 ≈ 13%.
- Check for bias. If you're consistently under or over, adjust your seasonal factors or averaging window.
- Track stockouts and overstock. Count the days each A-item was out of stock, and list items with more than a set number of weeks of supply.
- Watch inventory turnover: annual cost of goods sold ÷ average inventory value. Rising turnover with few stockouts means forecasting is working. Our inventory turnover calculator does the math, and our explainer on what inventory turnover tells you covers how to read it.
Common mistakes
- Forecasting from sales during stockouts. If an item was out of stock for two weeks, sales were zero, but demand wasn't. Exclude those weeks or estimate lost sales, or you'll order too little next time.
- Ignoring lead time changes. If a supplier's delivery time stretches from two weeks to four, your reorder point needs to roughly double. Update lead times whenever they change.
- Chasing volume discounts. A 5% discount isn't worth it if the extra stock sits for months or ends up marked down. Compare the savings with the cash tied up and the risk of leftovers.
- Treating every item the same. Weekly reviews of every item are unsustainable. Use ABC classification.
- Forgetting promotions and events. A planned sale, a local event or a new product launch will move demand. Adjust the forecast for those weeks.
- Relying on a single supplier for key items. One late shipment can undo good forecasting. See why relying on one vendor is risky.
Where AI can help
AI tools can handle much of the number-crunching. Export 12 to 24 months of sales by item and week, and ask an AI assistant to calculate moving averages, seasonal factors and ABC classes, and flag items whose recent sales depart sharply from their usual pattern. It can also build a reorder sheet that applies your lead times and safety stock rules automatically. Our guide to using AI for inventory and demand forecasting goes further.
Review its output before placing orders. The tool doesn't know that a local race is coming, a supplier is discontinuing a product, or last spring's spike came from a one-time bulk order. Your knowledge of the business is what turns a forecast into a good order.
Frequently asked questions
How much safety stock should I keep?
Enough to cover typical spikes in demand and delivery delays. A simple starting point is one lead-time period or a week or two of forecast demand for important items. Adjust based on how often you actually run out or end up overstocked.
How far back should I look when forecasting?
Use recent weeks, often 8 to 13, for the base level, and a full year or more of history to estimate seasonal patterns. Very old data may not reflect your current customers or product mix.
How do I forecast a brand-new product?
Start with a small initial order, based on a comparable product's sales, and a short review cycle. Reorder quickly once you see real demand. Accepting a possible stockout early is usually cheaper than being stuck with excess.
Should I use a spreadsheet or inventory software?
A spreadsheet works for a few dozen key items. Inventory software becomes worthwhile when you manage hundreds of items, multiple locations or sales channels, or need automatic reorder alerts.
Key takeaways
- Forecast = recent average sales × seasonal adjustment.
- Reorder point = lead-time demand + safety stock. Order up to a level that covers lead time, the review period and safety stock.
- Round orders to case packs, and weigh volume discounts against cash tied up and markdown risk.
- Use ABC classification to focus effort on the items that matter most.
- Track forecast error, stockouts and turnover to improve over time.
- Exclude stockout periods from history, and update lead times as they change.
Comments
Post a Comment