You do not need a forecasting model to make a sensible estimate of next month's sales. For a small catalog, three numbers from your sales history and some multiplication will do. This guide walks through the method and where it breaks down.
Start with a baseline
Your baseline is how many units of a product you sell on a normal day right now. Take a recent period, such as the last 30 days, and divide units sold by the number of days. Shopify's inventory reports include one called Inventory sold daily by product, which shows the average number of units sold per day for each variant over the period you select.
- Use a recent window: Recent sales say more about next month than sales from a year ago
- Check for stockouts: Days with nothing to sell pull the average down, so the true rate is higher
- Check for promotions: A sale week pushes the average up, so the true rate is lower
Adjust for trend
Next, check whether sales are rising or falling. Compare the last 30 days with the 30 days before that. If the product is growing or shrinking steadily, carry part of that change forward.
- Flat: Keep the baseline as it is
- Rising: Raise the baseline by the growth you saw
- Falling: Lower it, and be cautious about large reorders
Adjust for season
Then ask whether the coming period is normally stronger or weaker than the one you measured. Look at the same two periods last year and divide one by the other. That ratio is your seasonal factor.
- Factor above 1: The coming period is normally busier
- Factor below 1: The coming period is normally quieter
- No history from last year: Use a similar product, or skip this step and hold more buffer
Put it together
Multiply the three numbers by the number of days you are planning for. Here is an example for a 30-day forecast.
| Step | Number |
|---|---|
| Baseline, last 30 days | 6 units a day |
| Trend, up from 5 a day the month before | × 1.2 = 7.2 a day |
| Season, last year this month sold 1.5 times the month before | × 1.5 = 10.8 a day |
| Forecast for 30 days | 10.8 × 30 = 324 units |
Treat the result as an estimate, not a fact. Trend and season can overlap: if part of last month's rise was the season starting early, applying both counts it twice. When you are unsure, use the smaller adjustment and review after two weeks.
Where the manual method runs out
This works when you have a short product list and time to repeat it. It gets harder as the catalog grows, because each variant needs its own baseline, trend and seasonal factor, and each one goes stale.
- ✅ A few dozen products with steady sales
- ✅ A monthly review you will actually do
- ❌ Hundreds of variants, each with different patterns
- ❌ New products with no history to measure
Once you have a forecast, feed the daily rate into your reorder point so the estimate turns into an order date.
Doing this at scale
StockSense forecasts per-SKU demand from your order history using velocity, trend, seasonality and lead-time-aware safety stock. It recommends a reorder point, quantity and date for each SKU, and ranks SKUs by revenue at risk. See StockSense.