Product taxonomy is vital for any eCommerce company. A distributor has no choice but to optimize the classification and organization of its products, both for eCommerce success and other facets of business operations. Unfortunately, companies face numerous obstacles to getting their product hierarchies right, resulting in inaccurate classifications and confusing organization.
The good news is that solutions to these product classification woes are now available, in the form of AI and machine learning tools. These tools can apply a semantic approach to placing products in hierarchies. With these AI solutions, your product taxonomy can stop being a headache and instead set you up for eCommerce success.
Product Taxonomies are Mission Critical for eCommerce and Beyond
There is one obvious area where product taxonomy is essential for distributors selling online — the eCommerce website itself. Customers must be able to navigate your website easily, with minimal friction or confusion. This means having your products well-organized and classified correctly, which will help your customers find what they are looking for in the fewest clicks possible — ultimately leading to increased sales.
Beyond the website, the way a distributor organizes its products is critical for every way the business is run. Consider just a few of the numerous business impacts of the company’s product classification:
- Analytics: Many business analytics and reporting, such as year-end profit-and-loss statements, contain information divided between product families.
- Product Management: Product managers in charge of different product families rely on the product taxonomy to outline their responsibilities.
- Vendor Management: Effective vendor management often requires a distributor to classify its suppliers, and the supplier classification will largely be based on the product hierarchy.
- Responses to Tenders/Bids/RFQs: When receiving requests for bids, tenders, or RFQs, distributors must align the incoming parts of the bids with their internal product hierarchies so they can accurately identify pricing and strategize their responses.
Product classification affects many of the inner workings of the company, which means poor classification will hamper business processes in multiple ways. In short, product classification is something that distributors cannot afford to get wrong.
Barriers to Optimizing Product Taxonomy
If optimizing product taxonomy is so vital, why can’t distributors consistently accomplish this task? What is standing in the way of product classification success? There are several culprits.
Manual Product Classification is Ineffective
Manual product classification is not effective for modern eCommerce. Manually categorizing products one by one is a time-consuming process, with great potential for errors along the way. In addition, assigning items into categories can be a subjective process — it is difficult for all personnel to follow the same standards of categorization.
In modern eCommerce, the high flow of inbound product information cannot be matched with manual classification. Most distributors have vast product catalogs and hundreds, if not thousands, of suppliers. Manual processes cannot keep up with this complexity and volume. Accordingly, products are not accurately classified and loaded onto the eCommerce site at the rate they need to be, and the distributor misses out on sales volume.
More Hierarchies, More Problems
Adding to the difficulty, many companies have multiple product hierarchies. Each hierarchy needs to be managed, and the hierarchies must map to each other — yet another obstacle to effective classification with manual processes.
A company could have multiple hierarchies for various reasons. A distributor may sell products through multiple websites catering to different marketplaces, thus requiring a different hierarchy for each marketplace. Company acquisitions also result in the acquired companies’ hierarchies being added to the mix. Even different business units within the same company can have different standards for product categorization — and thus different hierarchies.
There is also a need to generate product hierarchies that conform to third-party standards for product classification. The commonly used third-party standards — such as the United Nations Standard Products and Services Code (UNSPC) or ETIM international standard for technical products — provide a common set of rules for identifying products. As simple and efficient numbering systems usable by both suppliers and customers, these standards are well-adapted for eCommerce. Distributors miss out on a competitive edge by not aligning their products with them.
AI Tools and Machine Learning to the Rescue
The good news is that solutions are available for product classification troubles. Artificial intelligence (AI) — and specifically the AI subfield known as machine learning (ML) — provides an excellent use case for product classification. The right AI tools can utilize machine learning models to categorize products quickly and accurately.
AI Tools Can Apply a Semantic Approach to Product Taxonomy
With AI tools that employ a semantic approach, users can teach machine learning models to interpret words and phrases to infer meaning. Using natural language processing and deep learning, these models can flag specific words or phrases from unstructured product descriptions and assess their meaning. The models then utilize the assessed meaning to assign the appropriate product category.
In addition to the product description, the AI applies the semantic approach to the product hierarchy itself. This enables the tool to match products to the appropriate branch of the hierarchy at run time.
The ability of an AI tool to understand the semantics of product descriptions and classification hierarchies is a true game-changer. The semantic understanding means the models do not need to rely on exact word matches or rules only. Instead, the machine learning model takes into account synonyms, variants, and the similarity of concepts and phrases. As such, classification is far more adaptive, consistent, and scalable than manual approaches.
It is important to remember these AI tools are not a magic wand — a poorly designed eCommerce website will still perform poorly, no matter the technology used. Accordingly, the best practices for product taxonomy still apply when using AI-powered solutions. You will still need to build your site around the typical user’s needs and research the competition. And you will need to avoid common no-nos in product taxonomy, such as placement of a product in more than one category, or mixing up product categories and product attributes. See the footnote below for more details on product taxonomy best practices. [1]
How AI Tools Can Assist with Product Classification
Powered by the WrangleWorks platform, modular transformations known as Wrangles can be utilized to meet your product classification needs. With the WrangelsXL Add-in, users can create, update, and share Wrangles themselves, then apply them to selected data from Excel spreadsheets with just a few button clicks.
A Wrangle can assign products to categories based on the semantics of the product description. In addition, Wrangles can provide a confidence score showing the level of certainty that the classification is correct. This allows users to streamline the process by making strategic choices, such as only double-checking classifications below a certain confidence threshold — 90% or 95% confidence, for example.
With an innovation unique to WrangleWorks, Wrangles can also suggest how to categorize your products, leaving the ultimate choice up to the user. The Wrangle will show potential choices for the product category, along with confidence scores for those choices. This provides the human behind the keyboard with their own level of confidence that the models are working well for classification purposes.
AI tools such as Wrangles can also match between one hierarchy and another. This will allow a distributor to match a supplier’s taxonomy to its own internal taxonomy, then correctly classify the supplier’s products. This function is also valuable for companies with multiple internal hierarchies.
It is important to remember that this will not be an overnight event. Even with the help of Wrangles, the product classification process is more of a journey. It will take time to analyze the product taxonomy, make categorization decisions, and make ongoing refinements. Nonetheless, Wrangles can use semantics to make the classification journey proceed more quickly, with enhancements not otherwise possible.
Handle Your Product Taxonomy Headaches with the Best AI Tools for the Job
Your product classification efforts do not need to be held back by manual workflows. Instead, AI and machine learning can revolutionize the classification process. AI-powered tools, such as Wrangles for eCommerce, can apply a semantic approach to optimize your product hierarchies in a fraction of the time required by manual efforts — helping alleviate your product taxonomy headaches.
[1] For B2B product taxonomy best practices, see Your Guide to Product Taxonomy | Bloomreach and
Build an Effective B2B Product Taxonomy: 7 Best Practices (vservesolution.com). For B2C best practices that are still applicable in the B2B world, see 7 Steps To Successful Product Taxonomy - Gepard PIM and eCommerce Product Taxonomy Best Practices for Merchandising Excellence (crobox.com),
