This is just some announcement which people will want to pay attention to! Learn More
This is just some announcement which people will want to pay attention to! Learn More

Attribute-based demand forecasting

Most brands already take advantage of demand forecasting in an attempt to more accurately determine how much of their products to produce and when. Their tools combine product SKUs with past transaction data to predict future demand from various retailers and distributors. This technique can be broadly effective: producers can make relatively accurate predictions about existing products and how many shipments of them will be needed monthly or weekly. If the business can trust their forecasting, it can help them make better decisions about staffing, resources, and investments. 

However, this simple means of demand forecasting can leave crucial—and expensive—questions unanswered.

  • Why are Nike Air Max in size 8 selling out at twice the rate of the same shoe in a size 7.5?
  • Which color Nike Air Max is most popular with female purchasers?
  • What is the difference in popularity between a 10.5 oz box of Honey Nut Cheerios and a 22 oz box of the same cereal?
  • What is the difference in popularity between a 22 oz box of Honey Nut Cheerios without promotional branding, and the same size box with promotional branding from Pixar?
  • How much demand is there for gluten-free organic cereal options versus simply gluten-free cereal options?
  • How much more likely are customers to purchase a potential new Mountain Fresh flavor of mouthwash over a flavor like Wintergreen that has been offered for several years?

The possibilities are endless because product variations are endless. Unfortunately, in current demand forecasting configurations, all research and predictions are tied to product SKUs, and product variations often fall under the same SKU. If strict GS1 standards are adhered to in designating GTINs to SKUs, every variation would be captured as a different GTIN but this often isn’t the case. A single SKU often encompasses multiple flavors, packaging sizes and appearances, and item sizes and colors. And because there is no historical transaction data for new products, they cannot be accounted for in any research or predictive models. 

Attribute-based demand forecasting from CrowdANALYTIX is different. It incorporates as many attributes of a product as possible, giving businesses the ability to separate these attributes from the SKU. A company can slice and dice their products on demand, and answer much more detailed questions about demand. 

Figure 1: Extracting product attributes from images and pdfs.

CrowdANALYTIX’ approach involves four unique modules that work together:

  • Module 1: A suite of hundreds of models is created to extract product attributes consistently, from images, PDFs, and any other file type. 
  • Module 2: Elasticity models by product type are made to determine sensitivity to pricing changes (needed for price promotion applications only). 
  • Module 3: Clustering models are established to identify products that perform similarly and influence each other, to determine the halo effect of SKUs. 
  • Module 4: A forecasting model is created to combine and leverage the other three modules.

The result is an intuitive dashboard that allows users to search for performance according to myriad product attributes. For example, in the GIF above, the company can determine how much more successful the Peppa Pig packaging makes their product, which flavor of toothpaste is most popular, how important the “gluten-free” designation is to consumers, and much more. Brands can gain new insights they never had access to before, resulting in more successful new product development, better product and packaging design, and far more informed decisions about resource purchasing and staffing.

The insights provided by attribute-based demand forecasting can be extended to the practice of store assortment optimization. This means predicting not only how much of a product to produce and ship, but also where exactly to ship it. Using the dashboard, businesses can drill down into their product assortments and determine exactly which stores sell the most of a particular product. For example, if a brand distributes their products to WalMart, they can determine which WalMart locations will be best to try out a new product packaging, flavor, or color; which locations need extra items during the holidays; and exactly when each location will need a new shipment. Accurately answering these questions can increase sales, reduce waste, and make it easier for brands to successfully establish new products and product versions. 

Share This Post

Share on facebook
Share on linkedin
Share on twitter
Share on email

More To Explore