Automate Product Classification For eCommerce Filtering & Search

Posted by Maria West on Jul 5, 2020 6:58:27 PM

It’s well accepted that on-site search and filtering is important. The product taxonomy help’s visitors narrow in on the specific products they’re looking for and find products they might be interested in. This is particularly critical with large product catalogs, for visitors to enjoy and appreciate the extensive range instead of bouncing in frustration. 

Interestingly, despite the logical importance of filters, The Baymard Institute usability testing across 50 major eCommerce sites found that 84% of the sites had a mediocre or poor filtering experience. 

So why, given the advantages of filtering, are so many eCommerce sites falling short?


Poor product taxonomy and classification

The answer, we see with the retailers we work with, is not a challenge to build filter functionality on the website. It is the exasperating amount of time needed to manage the product classification that feeds the filter logic. 

Retailers can work with their eCommerce agency partners to design a good data taxonomy to feed the filters. But the product catalogs they receive from their suppliers, are all in different formats and structures to that architecture. Cue hours and hours of work in excel for the retailer, to translate each vendor's product data into the format needed ready for their website.  

For example, a useful filter might be product length. Different suppliers will provide that attribute in different units of measure, such as cm, in, m or ft.  Or different formats for the same unit of measure, such as Length: 10IN, another Length: 10” and the last Length: 10, Length Unit: Inch.


Modmart measurement filters


For an easy to use filter and shopping experience, those measurements will need to be reworked into a consistent format.  

The large amount of work needed to re-work this data often leads retailers to make uncomfortable choices such as

1/not including filters on the site

2/including less filter options, or

3/ limiting their range size as a whole to a size they have the time to manage. 


Automate product attribute classification with Vesta eCommerce

The answer to this challenge is using automation tools to transform the incoming supplier data into the right format. Cutting out the heavy lifting of manual manipulation work in spreadsheets. 

Vesta works with the retailer to do this, by setting up rules for each incoming supplier feed. Once this initial setup is complete - all future supplier catalog updates, automatically run through Vesta's product data transformation engine, outputting the data in the correct data architecture for the eCommerce site. 

Implementing this solution, removes the barriers holding retailers back from implementing solid filters on their site, and greatly increasing their catalog size. 


Here are some examples of the types of transformations easily managed through Vesta:


Standardized attribute options

Take a wide range of supplier attribute names and funnel them into a concise, standardized set of filter choices. 

Eg. Supplier color names - ‘navy’ ‘indigo’ and ‘soft sky blue’, are programmed to map to the retailers appropriate color filter option - ‘’Blue’’

Modmart blue filter


Automatically generate Options Sets

Also known as Parent/Child, Simple/Configurable or Product Variants.  This is an area that often takes hundreds of hours of retailers time. 

Vendors often provide data in a “flat product structure”. A spreadsheet with 1 row per product, and oftentimes no information on how these individual SKU’s should combine to create a single product with multiple variants.

With smart automation through Vesta's Variant Builder, important Options can be extracted from your product titles and converted into a Parent/Child relationships. For example:

Supplied Data:

Ladies Polo 10 Grey/Blue

Ladies Polo 12 Grey/Blue

Black/Red Ladies Polo 10

Black/Red Ladies Polo 12


Vesta can transform it to:

Parent: Ladies Polo


Colour Options: Grey/Blue, Black/Red, Black/White, Blue/Green

Size Options: 10,12,14, 16,18


Let's not even start on when the size 16 can't be purchased in Black/White, but don't worry we have you covered for that too.

Category classification 

Categories are a key search and filter criteria for shoppers to find what they are looking for. Vesta can help to assign products to categories through a smart system of keyword checks. 

Eg If product title contains ‘’cap’’ sort into category ‘’Apparel/Headwear/Tactical Caps’’

HOWEVER if product title contains “White Sox”+”Cap” sort into category Apparel/Headwear/Baseball Caps

Expand Supplier acronyms

Vendor data is often full of acronyms and ''manufacturer speak'' which isn't easy to understand for the end consumer. Vesta can expand acronyms to full words and consumer friendly language for use in filters, and to enhance search


Eg BL = Belts; FL = Flashlights


Take this another layer deeper with Vendor specific acronyms e.g

If supplier equals “Vendor1” BL = Belt

If supplier equals “Vendor2” BL = Black 


Sort Orders

Specify sort order for different options within a filter to either put most useful ones at the top, or organize them into a logical order, 

eg. for clothing sizes - sort them in ascending order S - M - L - XL

What is the impact of automation tools to transform supplier product data?

As one use case example, a customer with a range in the hundreds of thousands, had an objective to deliver an easy intuitive shopping experience for their visitors. A key to meeting this goal was the use of product category and attribute filters. Over 8 months, Vesta has run 3,129,548 transformations over new products and catalog updates, saving them an estimated 26,000 hours or 13 years of one person continually working on manually manipulating the data in spreadsheets. This equates to $627,000 in labor costs. A saving that will continue to grow as we work together over time. 

Don’t let disorganized product data hold back eCommerce growth

Automation tools like Vesta are a necessary step to eliminate the pain retailers are facing in getting supplier data into the right format for their online stores.  In order to implement important features such as on-site filtering and search, product data must be classified into a well structured data architecture. For many retailers, this is overwhelming and they miss a big opportunity in sales due to not knowing how to solve these product data challenges. With Vesta, this doesn't have to be the case. If you are struggling with anything covered above, get in touch to chat to one of our friendly team and book a demo today

Topics: seo, eCommerce filtering and search, Product taxonomy, Product classification

Recent Posts