Sales forecasting for small businesses
How to use past sales data to plan stock, staff and cash flow, from simple methods to machine learning.
EvoAstra team14 Nov 2026 · 3 min read- Even simple forecasts beat guessing
- Account for seasonality and festivals
- Compare forecasts with actual results and improve
Forecasting sales helps you order the right stock, plan staff and manage cash. It does not need to be complicated to be useful.
Start with your history
Gather at least a year or two of sales data by day, week or month, ideally by product or category. More history helps the forecast understand seasons.
Simple methods first
- Same period last year, adjusted for growth
- Moving averages to smooth out noise
- Seasonal patterns around festivals and sale events
When to use machine learning
If you sell many products, have lots of history and many influencing factors such as promotions, prices and weather, machine-learning models can improve accuracy. They need clean data and regular checking.
Include what you know
Planned promotions, new product launches, price changes and marketplace sale dates should be built into the forecast.
- 1Collect history
- 2Pick a method
- 3Add known events
- 4Forecast
- 5Compare with actuals
Review monthly
Compare forecast with actual sales, understand the gaps and adjust. Forecasts get better with feedback.
Tell us what you’re working on. Free estimate within 24 hours, NDA on request.