The eCommerce merchandisers guide to building the best product taxonomies using product data.
The eCommerce merchandisers guide to building the best product taxonomies using product data.
7:00 am Monday morning. It was with shock and horror that I realized I was out of coffee. I dashed for my nearest supermarket. Shuffled through the store in my PJs. Made a beeline straight for the coffee section...only to find myself looking up at diapers.
Granted, I was a little groggy.
But the retailer changed how the store was organized overnight. What was once a subconscious journey through my favorite store (and to my favorite aisle), now jolted me back into a coffee-less reality.
I had to navigate around the supermarket to Breakfast - Cereals before I finally found the Teas- Coffee section.
When retailers reorganize their stores into hierarchical categories, this is usually done using a product taxonomy.
And - you’ve guessed it - product taxonomies are so important because they trigger the subconscious ease that’s felt from seamless customer journeys.
In eCommerce, having a strong product taxonomy is the crux of your digital merchandising and product discovery strategy.
With that in mind (and coffee in my belly), I’ll take you through product taxonomies from A-Z. There will be plenty of examples along the way so you can see taxonomies at work.
So buckle up, drink your coffee, and let’s get started.
A product taxonomy is the organization of products in a store. For eCommerce, this is your product categories and subcategories organized in a hierarchical structure (i.e., product hierarchy).
Product taxonomies are structured top-down, meaning you start from a high-importance category or page and work your way down, with branches that shouldn’t link back to the top. In other words, the structure isn’t cyclical.
“The bad news is that there is no unified approach to product taxonomies for fashion. The good news is that fashion retailers can create multiple taxonomies based on a single database.” -Leonard Wolters, CDO at Crobox
These categories shouldn’t be made up on the spot. Product taxonomies only work if you have coordination between your inventory teams, developers, SEO specialists, and PIMs (or product data coordinators).
House of Slippers taxonomy is horizontal and shows images of their categories.
Zalando’s taxonomy is a vertical bar on the left-hand side, with clear filtering options on the PLP.
You’ve probably already lamented how there’s no industry standard for taxonomies in retail. Even direct competitors like Adidas and Nike will use completely different systems of organizing their products.
Basically, it’s like going to the library. Every library has a different system of organizing books, from Dewey decimals to author names.
Why does this matter?
Well, for retail, this makes sharing product information across departments all that more important. Once you can share this information, it makes it so much easier to have a strategic approach to your product taxonomies.
Organizing your eCommerce products is important for a whole host of reasons.
“Product taxonomies are important because they bring orchestration and organization to a huge inventory of products.” -Leonard Wolters, CDO at Crobox
A good product taxonomy:
Without an organized webshop, customers will have to use their System 2 way of thinking. According to behavioral economist Kahneman, System 2 is the conscious, reflective, and deliberate side of the brain when it comes to decision-making.
What merchandisers really want to tap into is System 1: These subconscious, effortless, and quick reaction times.
Why? Your customers will leave a site in fifteen seconds if their journeys are stunted in some way, or the webshop doesn’t deliver on familiar UX.
More organization = elevated product discovery.
Because the path to finding and exploring products online is all about the decision-making journey your customers undergo.
If you can provide an easy, organized, and clean taxonomy, your shoppers won’t have to think too hard about where they’re going.
Unless, of course, they have a specific product in mind.
If your customers have a specific product or keyword in mind, they’re more likely to use your search bar, right?
A search engine optimized taxonomy will facilitate product discovery, both within your website and across the web. The best webshop designs go hand in hand with the best taxonomies, and can easily turn browsers into buyers.
Your product taxonomies also reveal a lot about your brand. For example, it doesn’t cut it anymore for brands to be apolitical, especially in the wake of COVID-19.
In fact, according to Edelman’s Report, 57% of consumers will boycott a brand because of their political or social stance. This means that it’s more important than ever to have your core values or CSR echoed from your webshop (and taxonomies) to your marketing, and across your omnichannel campaigns.
Brands that don’t capitalize on the opportunity to shine their tone of voice using their taxonomies will miss out on an important facet of merchandising engagement.
Source: Oribi
This last one is important for customer segmentation and general webshop optimization. Understanding how your shoppers navigate your product taxonomies is also important if you want to improve it and refine your eCommerce strategy.
Since every taxonomy category is tagged with a value (e.g.,“Male”), this value should represent a part of your customers’ contextual data. Once you know more about your customers’ contexts as they browse your webshop, this will help you collect that valuable behavioral data.
Collecting behavioral data about your taxonomy can be done by:
Or, you could use AI. Here’s where it gets interesting.
At Crobox, we use AI to continuously test different product attributes on the PLP and PDP. Following these experiments, we can make data-backed assumptions about which product features drive click behavior. This kind of behavioral data can also be reused to optimize your product taxonomy. 
For example, if we know the product badge “100% Recycled Down” drives a lot of interest (measured by clicks, for example), then it might be worth creating a separate category on your product taxonomy for sustainable products.
Our AI can test multiple messages at the same time, to see which copy-variant (and, therefore, which product attribute) appeals to which customers.
This is behavioral data that can then be reused for segmentation, product promotion and creation, and even your omnichannel campaigns.
Our product insights give you an understanding of the who, what, when, and why of your product engagement. Learning more about your products will facilitate the search and discovery experience - and AI is a good way to create flexibility in this process.
With all that in mind, let’s discuss some of the Do’s and Don’ts of creating a product taxonomy, so you have a better idea of how to apply these tricks of the trade yourselves.
Sharing product data across departments requires an un-siloed approach. I know many of you are suffering from silos, departmental or data-wise, which makes it harder to get a product taxonomy strategy up and running.
In short, therefore, your first task is to centralize and share your product data, to make creating a taxonomy easier. Then you can reap the benefits of a strong product taxonomy.
Because, although there is no industry standard, there are still some best practices to help you get up and running smoothly.
Let’s take a look.
Online consumer behavior is completely different than offline, so your in-store product categories won’t be the same as they are on your webshop.
The best way to figure out how consumers navigate through your webshop - and the hierarchies of your product categories - is by testing on-site behavior. We’ve already discussed how you can collect behavioral data (AI, EEG, heat-maps, micro-conversions) but don’t forget to supplant this data with demographic and psychographic data too.
You can also leverage navigation patterns (and analyze the bottlenecks at every step of your funnel), conversion rates, time spent on your pages, links clicked, bounce rates, etc.
Product taxonomies are basically product search retrieval mechanisms. But you have to make your taxonomies search engine optimized, otherwise, this will negatively impact product search and discovery and leave your customers frustrated.
Optimizing your taxonomies includes adding additional content like images, Meta Titles, or Meta Descriptions that relate back to your target keyword.
Google wants to provide users with the right information they are looking for, at the right time. Optimizing your taxonomies by doing extensive keyword research into what terms your customers are looking for will go a long way.
Take Rawganique. They win first place on Google’s SERP for this keyword, but the real magic is how they embed this keyword in their product description. The best practice for SEO is thus to follow through with your target words without keyword stuffing.
(A keyword-stuffed URL will look something like this:
rawganique.com/shop-organic/organic-fleece/100%-organic
Which you should avoid at all costs!)
You should aim to optimize your URLs with your keyword of choice in order to organize your products better and make product discovery easier for your customers.
You can also see what your customers are searching for by using a faceted search: Looking at how they filter through your page to refine their results (again, this is part of collecting behavioral data).
Adidas wants to be their customers go-to sports brand, and their extensive taxonomy reflects this as they organize their products by sports.
This product taxonomy best practice may seem straightforward but is something that many retailers are still lacking. While customer-centricity is all the rage, retailers are falling behind on understanding their products.
But if you can understand what product features are important to your customers - e.g., what attributes drive behavior - then a product-driven customer-centric approach will fall easily into your lap.
TwoThirds' product category includes stories around their eco materials because their products go hand in hand with their environmental pledge.
So how do you understand your products on a deeper level to facilitate your product taxonomy strategy?
You should ask yourself:
Burberry has a separate category for their famous Trench Coat - the product that is the face of their brand, and which still appeals to a large part of their consumers.
A product-centric approach on-site will help you continuously optimize your product datasets. And with strong datasets comes stronger taxonomies - quote me if you like.
The more structured your data, the stronger your taxonomies, and the better your product discovery will be on-site.
Nike product taxonomy in AE vs. the UK
Every country or culture will have a different way their consumers browse a webshop. That’s why customer segmentation is so important. Because of seasonality in the UK, Nike includes a “Winter Wear” category in their taxonomy, which isn’t included in the Arab Emirates webshop, for obvious reasons.
In short, you need to optimize your taxonomy given your audience.
“We’ve seen with our own clients that taxonomies differ per country. In most cases, you can’t have the same taxonomy in France as you will in the UK. It’s thus important to keep optimizing per geographic location, or your taxonomies will cause friction in the buyer’s journey.” -Leonard Wolters, CDO at Crobox
However, optimizing your taxonomy doesn’t work unless you are continuously testing and adapting. If you do a big reorg of your taxonomy, for example, you should test your navigation before going live.
At Crobox, we’re big fans of a culture of experimentation. We collect in-session data (that allows us to stay within GDPR and cope with ITP) from our tests that we feedback to our retail clients.
This data is in real-time and can help retailers understand how their users are behaving on their webshops.
ASICS Shoe Finder tool
For example, we helped ASICS optimize their Shoe Finder tool. Coupled with research from Braingineers, we found that many ASICS customers were getting distracted by “Trail” running options on the Shoe Finder quiz wizard.
Based on these insights, ASICS created a whole new category for “Trail Running” in their taxonomy. To really see why your products appeal to your customers, it’s all about testing to find the right attributes you want to display as categories.
Your customers aren’t searching for ‘other’, so you simply shouldn’t use this as a category. Instead, refer to your keyword research and products to create sub-categories that make more sense.
Have your products as the central point for product discovery. What your customers search for (e.g., "Brown, Leather-strapped watch") is probably how you want to organize your categories.
A product can’t be in more than one category because it will mess up your product reporting, stock management, and purchase orders.
Of course, products that are also in SALE or discount categories are fine. If you do have SALE/ discounted products that have badges attached to them, we recommend having a filtering option so that when customers click on the SALE badge they are directed straight to the SALE category.
You should also avoid duplicating your categories, as this will just look messy. Plus, you want to avoid too many choices (to decrease Choice Overload).
More than fifteen categories is probably not a great idea. Don’t forget product taxonomies are about triggering System 1 to streamline your customer journeys as much as possible.
So, instead of having the taxonomy:
You should have something that looks like this instead:
Ever played the game “Guess Who?”? When making decisions for your product taxonomy, you can think of the design of this game: 
Your opponent would ask a series of questions to know which character you chose on the other side, and by process of elimination would aim to figure it out.
The same can be done when crafting a decision tree for your product taxonomies by looking at your product data.
Source: Icecat (a product data company that reached out to manufacturers to centralize all product information).
For example, let’s say Adidas wants to create a subcategory for their Men’s sports tights. Within this category, they’d have separate branches for either color, size, or attribute. Their hierarchy would look something like this:
However, if Adidas is only selling one sports tights product for men, and fifty for women, then does it make sense to have a gendered split?
“You want to make use of the least amount of branches so that your process of elimination is faster. When creating a decision tree for your taxonomies, you need to balance the number of products per node, or branch.” - Leonard Wolters, CDO at Crobox.
For the technical ones amongst you, this process is called an Information Gain and is used to train decision trees in machine learning.
And who knows? The future of product taxonomies will most likely be automated, where ML will read your product data-sets and create taxonomies that make the most sense.
But for now, let’s stick to what we know.
The goal of a strong product taxonomy is so that your customers can find exactly what they’re looking through with ease. Psychologically, product taxonomies facilitate decision-making, which will instantly keep your customers coming back to shop with you.
The best product taxonomies will create multiple taxonomies from their product data set. They will test their organizational hierarchy with their customers to see what works and what doesn’t, and optimize accordingly.
In short, product taxonomies in eCommerce are supposed to be dynamic, product-centric, and user-friendly.
Now it’s up to you!
