Imagine taking on the vast and tedious task of categorizing and assigning searchable tags to millions of varieties of nuts, bolts, screws, and other products for a B2B e-commerce site—and doing it all manually. It would take an extraordinary amount of time, and you’d likely make more than a few mistakes along the way. Yet this is the status quo for many B2B companies, who often pay contractors large sums of money to parse through hefty product catalogs and attempt to make them more searchable. In many situations, this process is simply far too convoluted and laborious.
Non-automated product categorization is cost prohibitive, and there are more than just the up-front costs to consider. If a product is tagged inaccurately, it may not appear in a buyer’s search. Oftentimes the customer ends up disappointed, and the company loses out on a sale. For B2B companies like auto parts distributors that sell millions of products, this is a losing proposition.
To compete successfully in omnichannel e-commerce, companies need a better product tagging solution to enable more accurate search results—searches that allow SKUs to be seen under as many relevant search queries as possible. Enter automated text classification, a process that uses artificial intelligence and machine learning to place products into different categories based on their text descriptions. This not only saves time and effort, it improves the search experience, increases online sales, and helps B2B companies outpace their competitors.
The Status Quo is Slow and Expensive
Properly tagging and categorizing a broad range of products and making them searchable can be a huge challenge. It’s an especially complex problem for companies like General Electric, which sells millions of parts designed for specific end products.
GE’s product database contains an immense number of screws, tubes, and other components, many with slight variations, designed to fit specific refrigerators or washing machines. Even the smallest component must be categorized under the correct serial number of an appliance. Miscategorizing even one of those parts can be disastrous if it fails to show up correctly in a search.
This can be particularly problematic for distributors and original equipment manufacturers, or OEMs, many of which don’t manufacture the products they sell. Such companies have to collect, clean up, and categorize disorganized product information provided by their manufacturing suppliers, which is costly and time-consuming. And, once all of that product information is complete and accurate, the data needs to be converted to a format that’s optimized for complex B2B searches. In today’s fast-paced economy, companies can no longer afford the time and money it takes to do this all without the aid of machine learning and artificial intelligence, or AI.
Artificial Intelligence is Changing the Game
I predict that B2B companies will eventually completely automate product classification, particularly as AI and machine learning technologies become more sophisticated. As Jeff Bezos wrote in his annual shareholders letter, “Over the past decades, computers have broadly automated tasks that programmers could describe with clear rules and algorithms. Modern machine learning techniques now allow us to do the same for tasks where describing the precise rules is much harder.”
Auto-classification, or automated smart categorization of products, is based on an image and/or short description of the product. Auto-classification offers a silver bullet for companies that deal with thousands or even millions of products in their B2B catalogs. Using advanced machine learning and AI techniques like CNN, LSTM and GRU, algorithms are trained across thousands of categories and subcategories to identify other algorithms with a level of precision that consistently exceeds 90% or higher. Once the algorithms are properly fine-tuned, B2B e-commerce catalogs can be categorized at rapid speeds and at a much higher quality level than if humans were involved.
Classifying products into standard categories can transform the customer experience by making products far easier to search, which is essential for end users. According to Forrester Research, 90% of B2B buyers start their purchases with search, and 74% conduct half of their research online before making offline purchases. Some B2B companies take the search experience a step further: Atlas RFID, for example, categorizes items into buckets, allowing customers to click and drill down based on their preferences.
Many B2B companies are only in the early stages of transitioning to e-commerce, but competition is heating up: U.S. B2B e-commerce is on track to hit $1.2 trillion by 2021, according to Forrester. In order to survive and thrive in our increasingly digital economy, all online merchants—especially those with massive product catalogs—need to embrace automation. The more time and money saved in auto-classifying products, the more resources be can put toward optimizing customers’ online buying experience.Favorite