Throughout my career in the software industry, I’ve been very focused on enterprise search and have been consistently asked about taxonomy by customers. While it’s generally a good thing to have good taxonomy, I’ve realized how critical it is to consider the costs and benefits of massive taxonomy projects.
What is taxonomy in 2019?
First, let’s define taxonomy in the context of your digital properties. Taxonomy is defined as a scheme of classification. In the knowledge management world, it refers to the set of metadata that will be added to the content that is made available to the end users. For instance, if I’m creating a Knowledge Base about fruits, I would want to create a taxonomy that adds different metadata for the color, the size, the number of calories and sugar content.
As more brands invest in their digital transformation, taxonomy quickly becomes a point of debate among internal support, product, digital and commerce teams. Everyone seems to have an opinion about the role taxonomy plays in carrying out your digital-first strategy.
Now let’s try to dispel a few myths about the need for taxonomies and when they should be created, updated and maintained – and why they might be slowing down your digital transformation.
Myth #1: I must have a well-defined taxonomy before doing anything.
We hear this one a lot in the context of online search and relevance, and unfortunately this mindset can delays projects. While taxonomy can increase relevance, a relevant search engine should be able to return the right content even if the taxonomy is not perfect. Advances in machine learning technology such as Coveo ART (Automatic Relevance Tuning) have been able to make relevance even better by associating search terms with content.
For example, an AI-powered platform will recognize and quickly detect patterns between seemingly unrelated searches — and then connect those dots by associating the relevant content the user is searching for.
Myth #2: Building a self-service strategy is futile until I have a taxonomy.
Although this is a normal concern, my experience has shown this to be incorrect. The sooner you can make content available for your customers or employees to consume, the sooner you can benefit from the returns. Put another way, the sooner you “turn on” AI-powered search, the faster machine learning models can be automatically created to start making associations for the best content to solve the need at hand.
Obviously, your self-service experience has to be relevant if you want to ensure that people adopt it and keep using it, hence the importance of deploying an AI-Powered Insight Engine like Coveo.
Myth #3: I need one unified taxonomy across all of my content sources.
Having a unified taxonomy across all content sources is a plus, but it often doesn’t need to be done in the original content source system. Instead, tools such as indexing pipeline extensions will allow you to create multiple mappings between the different system at indexing time, so you don’t have to modify it in the source system. It then allows you to quickly map the values between systems without having to rethink the whole taxonomy structure in all the source systems.
Also, by making the content available to a broader audience, you will be able to capture analytics and evaluate what type of content is actually moving the needle on Self-Service success and track with precision self-service success, case deflection, content gaps and click-through rates.
So why would you want to build a taxonomy?
There are definitely some areas where taxonomy can help and you should still consider taxonomy as a part of your projects.
For instance, if you want to secure your content in a very granular way, you will have to build your taxonomy, so that your Insight Engine can leverage it. We saw a lot of examples where Coveo customers would only want to show documentation for products that their customers actually own.. In this case, a taxonomy is required. Having said that, it’s always important to still ask the question if this will delay your self-service project and how critical is it for customers to see documentation about product they don’t own, perhaps opening it up can lead to some awareness and product upsells.
Another benefit of taxonomy is faceting, in order to build great and robust facets, you need consistent and accurate metadata. While facets are optional, it adds some ease of navigation and can help your users find the information more quickly.
To conclude, nothing is black or white and taxonomy is definitely an important part of knowledge management, but the cost of delaying your self-service deployment must also be considered.