Most inventory forecasting guides assume you have a data science team and a $200/month tool. This one does not. Here are three forecasting methods you can actually use — with the real formulas, a full worked example, and an honest explanation of when AI forecasting is worth paying for and when a spreadsheet is enough.
Inventory forecasting is predicting how much stock you will need in a future period so you can order the right amount at the right time. Done well it prevents two expensive problems simultaneously: stockouts on your bestsellers and dead stock on your slow movers.
Gut feel works when you have 10-15 products with stable, predictable demand. You know roughly how fast each one sells and you order more when it looks low. The problem is that gut feel does not scale — and it systematically fails in specific situations.
A product that normally sells 8/day suddenly sells 35/day after a social mention. Your gut feel says you have weeks of stock. Your actual days of stock just dropped from 25 to 6.
October sales of a summer product look low. Obviously. But if you forecast November orders based on October data, you will under-order a product that might have a Christmas spike.
You cannot mentally track 100 different products and their individual velocity trends. Gut feel aggregates everything into a rough impression that misses the outliers — which are usually the most important ones.
If your supplier takes 21 days, you need to predict demand 21 days into the future when you place an order. The further out you forecast, the more a structured method beats intuition.
You do not need complex software to forecast inventory. You need the right method for each product type. Here are the three that cover 95% of situations a Shopify seller faces.
Sold 80, 85, 90, 75, 85 units over last 5 weeks. Forecast = (80+85+90+75+85) ÷ 5 = 83 units next week.
Easy to calculate. Works for evergreen products with no strong trend.
Treats all weeks equally. Slow to react to demand changes.
Last 3 weeks: Week 1 = 60, Week 2 = 75, Week 3 = 90. Weights: 50%, 30%, 20%. Forecast = (90×0.5) + (75×0.3) + (60×0.2) = 45+22.5+12 = 79.5 units.
Responds faster to trends. Better for products gaining or losing momentum.
Slightly more complex. Still misses seasonal patterns.
Product averages 100 units/month. Last December sold 280 units — 2.8x average. This December forecast = 100 × 2.8 = 280 units.
Captures seasonal patterns that moving averages miss completely.
Requires at least 12 months of data to calculate a reliable seasonal index.
A linen summer dress. Sales peak in summer, drop in winter. Here is how each method performs across the year.
| Month | Actual Sales | Simple MA Forecast | Seasonal Index | Seasonal Forecast | Variance |
|---|---|---|---|---|---|
| Jan 25 | 42 | 42 | 0.70x | 42 | 0 |
| Feb 25 | 38 | 40 | 0.63x | 38 | 0 |
| Mar 25 | 55 | 45 | 0.92x | 55 | 0 |
| Apr 25 | 61 | 49 | 1.02x | 61 | 0 |
| May 25 | 78 | 55 | 1.30x | 78 | 0 |
| Jun 25 | 95 | 62 | 1.58x | 95 | 0 |
| Jul 25 | 88 | 65 | 1.47x | 88 | 0 |
| Aug 25 | 72 | 66 | 1.20x | 72 | 0 |
| Sep 25 | 51 | 65 | 0.85x | 51 | 0 |
| Oct 25 | 44 | 62 | 0.73x | 44 | 0 |
| Nov 25 | 40 | 60 | 0.67x | 40 | 0 |
| Dec 25 | 36 | 58 | 0.60x | 36 | 0 |
Not every product needs the same forecasting approach. Using a seasonal adjustment formula on an evergreen product wastes effort. Using a simple moving average on a highly seasonal product gives dangerously wrong answers.
AI forecasting is genuinely better than manual methods for most Shopify stores. But it is not always worth the cost, and the marketing around it tends to oversell the magic.
AI forecasting connects to your Shopify store via API and analyses your historical sales data to identify patterns — seasonality, trend direction, day-of-week variation, and velocity changes. It generates a predicted demand per product for a future window, with reasoning explaining why the prediction was made.
The practical advantage over a spreadsheet formula is not magic — it is scale and speed. AI can recalculate forecasts across 200 SKUs every time new sales data arrives. A spreadsheet requires manual effort every time. For stores with large catalogs and time-sensitive reorder decisions, that automation gap is where AI earns its cost.
Debnix uses Google Gemini AI to generate demand predictions per product — each with a risk level, estimated stockout date, and the reasoning behind the prediction. It connects directly to your Shopify store and recalculates as new sales data arrives. At $21.99/month it is the most affordable AI forecasting option available — the next tier up is Prediko at $49/month.
These mistakes consistently produce inaccurate forecasts regardless of which method you use.
A demand forecast tells you how much you will sell. That number feeds into the rest of your inventory system.
Inventory forecasting is the process of predicting how much stock you will need in a future period — typically the next 30, 60, or 90 days — based on historical sales data, seasonality, and demand trends. For Shopify stores, it answers the question: how much of each product should I order, and when? Good forecasting prevents both stockouts (running out) and overstock (buying too much).
Manual forecasting using simple moving averages is typically 50-65% accurate. Weighted moving averages with seasonal adjustment reach 65-75%. AI-powered forecasting that accounts for trends, seasonality, and demand signals reaches 75-90% accuracy for most products. No forecast is perfect — the goal is to be accurate enough to significantly reduce stockouts and overstock compared to gut-feel ordering.
Shopify Admin does not have demand forecasting. It shows your current stock levels and basic sales reports, but it does not predict future demand or tell you when to reorder. Since Stocky was discontinued in 2026, there is no native forecasting tool in the Shopify platform. Demand forecasting requires either a manual spreadsheet process or a third-party inventory app.
For simple moving averages, 30-90 days of sales data is enough. For weighted moving averages, 14-30 days gives good results. For seasonal forecasting, you need at least 12 months of data to calculate a reliable seasonal index. For new products with no sales history, use proxy forecasting from similar products and treat the first 8-12 weeks as a data collection period.
Forecasting predicts how much you will sell in a future period. Reorder points tell you when to place an order so stock arrives before you run out. They work together: your forecast tells you the expected demand over the next 30-60 days, and your reorder point (based on sales velocity and lead time) tells you the stock level that triggers an order. Good forecasting makes your reorder points more accurate because the velocity input is based on predicted future demand, not just historical average.
AI forecasting connects to your Shopify store via API and analyses your historical sales data to identify patterns — seasonality, trend direction, day-of-week patterns, and velocity changes. It then generates a predicted demand figure per product for a future window, with an explanation of why the prediction was made. Unlike manual methods, AI can process hundreds of SKUs simultaneously, update predictions as new sales data arrives, and catch subtle trend changes a spreadsheet formula would miss.
Debnix connects to your Shopify store and generates AI demand predictions per product — with risk levels, estimated stockout dates, and reasoning. No formulas to maintain, no spreadsheets to update. $21.99/mo, 30-day free trial.
Built solo.