Is it possible to be too exclusive?
In my time performing product taxonomy builds I’ve come across many different situations that challenged the concept that mutual exclusivity was absolute. Simply put, the level of detail and the specificity of the terminal nodes in the structure seemed like a maintenance nightmare if mutual exclusivity in categorization was held without exception. This begs the question: Can a product taxonomy ever not be mutually exclusive?
What is Mutual Exclusivity?
First, let’s define what mutual exclusivity is and why it’s so important. Mutual Exclusivity is the property of a categorization hierarchy that states that each terminal node must have a unique definition from every other terminal node. You can’t have a Drills terminal node as a sibling to a Battery-Powered Drills node, regardless of how far each other lies as siblings in the hierarchy. Every product can only have one terminal node it appropriately belongs to. It is often referred to as the one home concept.
There are multiple implications to this concept. The first is that the schema, or the set of attributes, applied to a single node must be significantly different from the schema of every other node. Having two nodes with exactly the same schema means that a product is just as likely to belong in either, which violates the “one home” concept. Attempts to define “significantly different” vary by product taxonomy application, but there should never be two nodes with exactly the same set of attributes in a product taxonomy.
Due to this concept, all parent nodes in the hierarchy must also be mutually exclusive. If two parent nodes share similar naming structures but only one holds the appropriate classification for a product, user confusion will abound. Therefore, the entire classification system through the entire hierarchy must be mutually exclusive in a product taxonomy.
This principle underlies all product taxonomy development, and exists for a few reasons. First off, if only a single node is appropriate for any one product than all products of the same classification will have the same attribute set applied to them, leading to consistency in ecommerce applications where comparison shopping is a functionality. Secondly, it avoids having to change a set of terminal node schemas because a duplicate node had a change that must be represented in all locations where a product can be appropriately classified, leading the maintenance problems over time. Lastly, it reduces the amount of time it takes for a user to classify a product, as there is only one true path where a product can be classified.
Is Mutual Exclusivity Absolute?
In a word, yes. Although rules about attribute stuffing (Adding attributable values into the names of categories) and empty drawers (The rule that says every parent category must have at least two siblings) can have exceptions in order to meet unique business requirements, mutual exclusivity cannot. The maintenance issues, user confusion, and overall degradation in the data quality derived from the collection and enrichment process that would result would be detrimental to the function and governance of the data. Therefore, mutual exclusivity should never have an exception to it.
There are consequences to this. First off, when dealing with service parts or assortments that contain many similar-yet-significantly different classifications, the number of categories required to maintain these differences will expand. Where attribute stuffing rules are meant to reduce the number of terminal nodes, mutual exclusivity expands that count. This means that truly mutually exclusive taxonomies will be larger.
It also requires the definition of “Significantly Different Schema” to be well defined in a governance process. Leaving this definition open to interpretation will result in a long-term deviation from the understood definition at the outset. To avoid long-term entropy in data, this is a necessity.
Lastly, it requires stricter adherence to a philosophy that does not come naturally to non-product taxonomists. There are very few taxonomists in the world that specialize in every type of taxonomy, but there are many that specialize in product taxonomy. Hiring the right people to fill the taxonomy and governance roles can avoid the same entropy a lack of definition of the required level of differentiation will cause.
Should Navigation Hierarchies Be Mutually Exclusive?
Navigation hierarchies serve a different purpose than product taxonomies, and by their very use case do not need to follow the rules of mutual exclusivity. They are polyhierarchical in nature, although they don’t have to be. Because they are used for findability, having one terminal node where a product belongs may create situations where the customer navigating that hierarchy cannot find the product that best fits their need. Having a single product in multiple locations is beneficial.
Here is where product taxonomy and website navigation intersect. A mutually exclusive product taxonomy can be mapped to align product assortments with any navigational structure desired. Having good mutual exclusivity rules and adherence actually makes the location of products in a web hierarchy easier to predict. If your goal is to entice customers into the right purchase the first time, it is best to map your mutually exclusive product taxonomy to multiple mappings on a web hierarchy.
TrailBreakers.AI Taxonomy Services
At TrailBreakers, we have decades of experience in understanding all the best practices of product taxonomy, and partner with companies that can perform data normalization and backfill at scale. If you need help with your next taxonomy project, please contact us at info@trailbreakers-ai.