Crisp uses advanced statistical models and your own sales order history data to provide you with a forecast. The accuracy of that forecast is going to depend on what you put into Crisp.
To start out, we recommend uploading at least a year's worth of data for each product and customer that you want to generate forecasts for. 3 to 5 years of data are ideal. Dense data, containing many data points at the daily or weekly level, will also help improve Crisp's ability to generate accurate forecasts for you.
Once your data is uploaded, there are several things you can do to further improve your account's unique forecasting algorithms:
- You can improve Crisp’s forecast accuracy by giving Crisp insight into the spikes and dips in your product’s historical demand. It’s as easy as telling Crisp about past events that impacted sales, like store openings and closings, promotions, and one-time events. This is where Crisp’s machine learning capability comes in- once you tell Crisp the context around your data, Crisp will learn how to better forecast for your product’s unique demand.
- Brand new products with little to no sales order history won't be left behind either: you can set up product relationships to tell Crisp to use data from older products to seed the forecast for the new product.
- Setting up automatic data ingestion will ensure that your Crisp account is always using the most relevant data to forecast for demand.
While forecast accuracy relies on both your data and the time you spend informing Crisp about your product's unique context, the outcome is worth it: accurate forecasts ensure that you always get the right amount of product to the right store at the right time.