Online shopping can be overwhelming because there are so many choices. It has become even more important that retailers ensure their digital experience helps customers find the things they're looking to buy.
URBN is a global lifestyle and fashion corporation with four major brands: Urban Outfitters, Anthropologie, Free People and Nuuly.
Global fashion & lifestyle brands
Effective product discovery is key
Daily digital sale across all 4 brands
URBN Search was an internal business tool, and the primary users of this product were our web producers across our four brands. The web producers were responsible for:
Before I joined URBN, I would've never expected any human involvement with search. While search algorithms are getting smarter, it can still result in irrelevant results being displayed to customers when paired with bad metadata. That's where the web producers come in. The web producers use the search vendor to configure the search algorithm to return relevant results.
Unfamiliarity with search and search algorithm
Unfamiliarity with fashion and lifestyle industry
URBNSearch needed to live within our merchandising tool, URBNCat
No design files for URBNCat existed, so I had to recreate all the assets
I was responsible for the design strategy and creation of design artifacts like wireframes and user flows. I designed the UI and interaction patterns
The research team had conducted interviews with the web producers and sent out an opportunity score survey to understand how web producers across brands utilized our search vendor and aspects of it they appreciated and would want to see in a similar tool.
I was able to go through some of their findings and these pull quotes stuck out to me.
Clearly, there were a lot of issues so I met with the web producers to identify any additional themes. They kept mentioning a lack of visibility and clarity into the mechanics of the current Search vendor.
The search vendor utilized these concept of "rules" that define particular search and site behaviors.
Problems with rules:
You want a search strategy to promote our highest selling dresses when people search for dresses? Create a rule for that.
You want to create a new summer dresses category with dresses that all have an average customer rating from 3-5 stars and have weekly sales of x dollars? Create a rule for that.
You want to define the filter options for this new rule you created so folks can filter by colors or price, or size? Create a rule for that.
I designed three features for URBNSearch. My goal with the designs was to improve clarity, reduce redundant work, embrace their flexibility, and give them the tools to troubleshoot any issues as easily and as quickly as possible. Two of those features are highlighted in this case study.
The first feature is Filters. It is exactly what it means. The filters you see on ecommerce platforms that help you refine a product list to find more relevant products. The web producers managed filters for all our brands on the search vendor and it had lots of challenges.
The web producers mentioned defining filters as a painful process with the current search vendor.
Problems with filter management:
This is also referred to as Relevance tuning. This enables web producers to further fine-tune relevancy of their results to optimize the experience for the customer and the business.
Imagine you have one app called Magicapp and another called Mysticapp. If someone were to use our search engine to look for "magicapp", which would they find?
From the above example, Mysticapp would be the first result if all fields were equal because the name Magicapp is present twice within two different fields. Without the ability to tune field weight, you would run into uncomfortable situations like this, which is not ideal.
The web producers managed relevancy tuning in our old search vendor, and they expressed a lot of challenges trying to understand how all the UI controls affected the outcome.
Problems with ranking profiles:
URBN Search started rolling out in August 2019. Early reports showed steady performance and slightly higher lift in Average Order Value for search result browsers. Even though I had to recreate the design artifacts, we could repurpose existing interaction patterns and design experiences that worked well with the web producers' mental model.
Search Exits reported to be down
Slight lift in Average Order Value
Results page view per search down
Time after search reported down
Search Depth reported down
Appreciation and direct feedback from users after launch
Turning constraints into opportunities
Thanks for stopping by. Check out another project.