Case Study: GS Shop

horizontal rule

The Partner

GS Shop ( is a global commerce leader. A subsidiary of GS Group, Korea’s 6th largest conglomerate, GS Shop has grown from its origins in TV-based shopping to become a multimedia shopping powerhouse, offering shopping on mobile devices, online, TV sets and through catalogs. Millions of shoppers visit GS Shop each day through their mobile apps and their online site at

By listening to customers’ opinions, providing individualized and participatory experiences, and providing the best products, GS Shop has attained the status of a top ecommerce player in the Korean market.

The Challenge

For many years GS Shop and other Korean ecommerce companies were able to enjoy strong growth as the Korean population shifted much of their shopping to online and mobile.

But today, to grow you need to take audience from other online competitors, by delivering experiences which are more appealing. Having the right products and right prices is just the first step. Figuring out how to design the product viewing experience—the page layout, the navigation, the pictures, the words, the colors,—well, that’s just hard.

Experimentation—testing and learning—is key to figuring out what will drive results. But, as large as GS Shop’s web traffic is, traditional A/B testing was not delivering results fast enough…and GS Shop had hundreds of ideas it wanted to test.

How could GS Shop break this logjam?

The Solution

Enter Evolv.

“When we first heard about Evolv,” says Jaiy Won Woo, vice president of Mobile Business Division for GS Shop, “we were excited, but skeptical. Testing dozens of potential improvements all at once sounded great, but how could Evolv and its AI be so much better than existing experimentation solutions?”

The GS Shop team decided to test Evolv’s optimization capabilities on its product detail pages. They identified 8 different areas of the page to work on and created 28 different variations to test, including:

  • Price Displays: How prices were presented to the shopper, including potential discounts
  • Call to Action Buttons: What colors and font sizes were most effective?
  • Layout changes: Which sections to emphasize and which to remove?

There were almost 5 million possible design combinations.

The GS Shop team used Evolv’s audience filtering and throttling systems to limit the shoppers that would see the test to only 10% of mobile users.

The goal of the experiment was set to maximize the percentage of shoppers that converted to a sale after viewing the product detail page.

The Results

In just 5 weeks of testing, even at only 10% of users tested, Evolv evolved 215 different design combinations, across 9 generations of evolution, to find the best converting design.

The winning design incorporated 9 out of the 28 potential improvements, and showed an increase of 6.4% in conversions.

Next Steps

Having seen how Evolv can help it achieve faster revenue growth more easily, GS Shop is now rolling out Evolv to all of its product and UX teams, and has begun a set of additional tests, with more on the way. GS Shop is using Evolv for both massively multivariate tests and also as an A/B testing solution where needed.

“GS Shop is focused on delivering users the best experiences possible for their shopping, wherever they shop,” noted Woo. “By harnessing powerful technology like Evolv, we give our UX, product and marketing teams more power to be creative and try new things, to increase satisfaction for our shoppers while further growing our business.”

By The Numbers

GS Shop harnessed the power of Evolv’s super-efficient optimization, which evolved the best converting designs out of the 5 million possible.


week test duration


elements tested




possible designs evaluated


more conversions

"By harnessing powerful technology like Evolv, we give our UX, product and marketing teams more power to be creative and try new things, to increase satisfaction for our shoppers while further growing our business."

– Jaiy Won Woo
VP of Mobile Business Division, GS Shop

Want to learn more?