“Unfindable” products can’t be purchased
Large retailers with extensive product catalogs face a problem that grows along with their rapidly expanding inventories: the presence of “unfindable” products. The cost of these unfindable items is high: many of them are in demand by customers both online and in-store, but if the products cannot be located through online search or by employees searching the catalog in-store, they cannot be purchased and potential income is lost.
What makes a product unfindable in a catalog or inventory search? The central problem is poor conventions of naming and tagging, combined with a constant influx of new products to the business. If the business is outsourcing its naming and tagging to human resources, it is almost impossible to keep up with growing inventory and the level of human error in data entry is relatively high. The result is that products are entered into the catalog with names and tags that do not lend themselves intuitively to search.
For example, an item like a ceiling fan had a simplistic naming convention that included only the brand and the item’s width. Unfortunately, customers were seeking ceiling fans using many more specific tags, including an indoor/outdoor designation, number of blades, color, etc.
Automation could be the answer
The retailer recognized that to improve their sales and sell these products that were “hidden” from search, they would need to expand product names and apply more tags to items. Most importantly, these new conventions would have to be applied not only to every new product being onboarded, but also to those already contained in the company’s huge inventory. The prospect of hiring enough human labor to accomplish these tasks by hand was daunting, and potentially impossible.
In their search for answers, the company began to consider AI automation to take over the process of editing and adding to the product catalog. With an automated process, the retailer could not only save money on their human outsourcing, but also speed the process of product onboarding without compromising on accuracy and thoroughness. AI could make it possible to save money while also improving searchability and therefore reducing the number of “unfindable” products hiding in the retailer’s massive inventory.
CrowdANALYTIX was chosen to attempt the automation process for this national chain of retail stores. In an initial analysis of the company’s data, CrowdANALYTIX found that some product types had “weak name” rates of up to 74%, meaning that 74% of products in a category were difficult for customers and employees to find.
After completing an analysis of the retailer’s internal data, including product data, category specific attributes (search filters), current taxonomy, and internal search results, CrowdANALYTIX could develop new, more effective naming and tagging conventions for the company. At least 50% of the retailer’s products were in need of significant updates. CrowdANALYTIX based their taxonomic recommendations on actual keywords customers were using to seek specific products both internally and externally.
CrowdANALYTIX projected that existing data could be enriched by 93% with the application of the customized DataX auto-tagging solution. Since CrowdANALYTIX used actual customer data to make their taxonomic recommendations, the impact of these changes was projected to be significant.