Back to Customers

How Curtsy Achieved a 10% Lift in Search Conversion Rate with Constructor

+10 %
Search Conversion Rate
Industry
Womenswear
Headquarters
San Francisco, California, USA
Solutions
Search, Recommendations
+10 %
Search Conversion Rate
Go to the Results

About Curtsy

Curtsy is a sustainable women’s fashion marketplace serving millions of U.S. shoppers and sellers.

“Instead of a buried feature, search is now something that we actually
want people to see — and we’re finding new ways to get them there.”


Dedicated to simplifying the thrifting process and promoting sustainable fashion over native and web applications, Curtsy has driven more than $36M in resale earnings on brands such as GAP, Madewell, Banana Republic, Free People, BCBG, and Urban Outfitters.

Go to the Results
Screenshot 2024-06-11 at 5.39.46 PM
Screenshot 2024-06-11 at 5.39.46 PM

“Instead of a buried feature, search is now something that we actually
want people to see — and we’re finding new ways to get them there.”


Dedicated to simplifying the thrifting process and promoting sustainable fashion over native and web applications, Curtsy has driven more than $36M in resale earnings on brands such as GAP, Madewell, Banana Republic, Free People, BCBG, and Urban Outfitters.

Go to the Results

The Challenge

With a highly rated native app across iOS and Android and a mission to provide best-in-classsupport, customer experience has always been top of mind for Curtsy.

But the functionality of Algolia’s product search didn’t live up to Curtsy’s standards—or their users’ expectation

 

Instead of driving users to search, the Curtsy team suppressed it in the app’s UI by eliminating a persistent search bar at the top (and prioritizing product feeds instead).


The Curtsy team applied machine learning models to intercept popular searches, bypass Algolia, and return their own optimized results. While this provided better outcomes, it became a highly manual process and required a lot of tech resources.

“Algolia’s text-based search is inherently flawed because it misses the user’s search intent."

“We’d just been willing to accept
that flaw and compensate by
deprioritizing search.”
Eli Allen
Founder & CTO, Curtsy
Go to the Results

The Solution

Where Constructor Comes In

A Better Ecommerce Search Solution

Once Curtsy got to the point where they had begun implementing their own temporary workarounds to the search experience, it was clear that improving search had become a priority.

 

“One of our big initiatives this year was actually showing people what they expect and want to see in front of them, rather than items based on popularity or text matching,” Allen says. “We were introduced to Constructor through an investor, and it was immediately interesting. A platform built around user intent was something that we had been talking about for a long time.” 

 

Working side-by-side with Curtsy, the Constructor team implemented the Proof Schedule to prove its ability to drive KPIs. “As soon as we had Constructor implemented,” Allen recalls, “we played around with it. We would do side by side comparisons of Algolia and Constructor results, and in most cases we were a lot happier with Constructor. The team was really happy about it, and when we showed it off to our investors, everybody was impressed.”

Constructor + Curtsy = All Stars


With Curtsy’s previous search solution, one persistent pain point came in the form of Converse shoes. Converse All Stars would not consistently appear when a user would enter a search query for “all stars.” Instead, Algolia’s text-based algorithm would return any item with “stars” in the product title or description—from jewelry to pants to tees. 

 

 

With Constructor’s machine
learning and natural language processing, results are displayed in part according to popular intent for each search term.

Today, a Curtsy search query for “all stars’’ provides a better customer experience that’s more in line with what the user is expecting. The result: happier shoppers and sellers, more click-throughs, and more conversions.

all stars results

With Constructor’s machine
learning and natural language processing, results are displayed in part according to popular intent for each search term.

Today, a Curtsy search query for “all stars’’ provides a better customer experience that’s more in line with what the user is expecting. The result: happier shoppers and sellers, more click-throughs, and more conversions.

all stars results

The Results

The positive reaction Constructor received from Curtsy’s internal team and investors was soon supported by quantitative data on key ecommerce metrics like conversion rate, reformulated searches, and clicks.

+23 %
in recommendations clicks vs Algolia
-14.5 %
in reformulated searches vs Algolia
+10.7 %
in search conversions vs Algolia
Curtsy was among the first to leverage Constructor’s Cognitive Embeddings Search, a new product feature that automatically returns the closest result for any search query and virtually eliminates zero results searches. For Curtsy, Cognitive Embeddings Search also eliminated the manual work that went into trying to provide better search suggestions for user queries both with and without search results.

Conclusion

What's Next for Curtsy

One way Curtsy has doubled down on search since moving to Constructor is introducing a “saved search” feature. 

 

Now a central part of the site’s UI, saved searches encourage users to revisit the site and perform more searches, not less. Allen says that this functionality just wasn’t possible before. 

“It’s only a good feature if the results are actually good,” he explains. “It used to be we were just actively trying to get people not to search, and now it’s quite the opposite.

 

Constructor has also shifted how the Curtsy team is able to manage online merchandising rules by exception. They found that they could trust and rely on the Constructor platform to return consistently relevant and attractive search results that drive conversions and revenue. All of the time the team had spent previously on optimizing and deprioritizing the search experience? “We don’t have to do that now,” Allen says. “We don’t have to design these complicated systems. And not only have we been able to reduce a lot of that complexity but we’re finding new opportunities to drive people to the search page.” 

Is there room for improvement with your current search vendor?​ Find out.

Reach out to a Constructor Product Discovery Specialist to learn how our testing helps surface gaps so you can make a data-driven assessment.

Book a Demo