B2C Retail Fashion

Popken boosts sales with smarter product listings personalized by Atlas AI

Key results

+4.3% conversion

Greater relevance thanks to real-time personalization with Atlas AI

Data-driven growth strategy

A/B testing with FactFinder accelerates digital evolution

Global efficiency, local relevance

Over 50 shops automated and easily managed

Meet Jason Ballard

Onsite Manager

Jason Ballard brings more than 10 years of eCommerce experience to his role as Onsite Manager at the Popken Fashion Group. Together with his team, he drives the ongoing development of the company’s online stores, with a strong focus on data-driven optimization of digital product discovery. That includes everything from site search and campaign control to personalized result sorting and product recommendations.

The Popken Fashion Group stands for stylish, size-inclusive fashion done right. The company’s four in-house brands, Ulla Popken, Laurasøn, Studio Untold and JP1880, are sold in over 250 retail locations and more than 50 online shops around the globe. As an omnichannel player, Popken offers trendy, high-quality fashion with great fit at a fair price. With core values rooted in customer centricity, innovation and responsibility, the brand continues to set new standards, especially in eCommerce and curvy fashion.

Challenge #1:
Legacy search system hindered global expansion

From the launch of its very first online shop, the Popken team knew product discovery would play a strategic role in customer satisfaction, revenue and operational efficiency. Initially, they opted for a well-known but rigid search solution that met their needs. That is, until global growth picked up speed.

As more international shops and languages were added, the performance of the old system deteriorated. Data exports lagged, indexing times stretched, and maintaining quality at scale became a serious bottleneck. The tipping point became the planned launch of the US shop. It was clear: Popken needed a more powerful, scalable product discovery solution that could keep pace with the company’s expansion.

Solution #1:
AI-powered product discovery that scales

To make the right call, the team evaluated several providers, placing performance as a top priority. Indexing times across all market-specific data sets had to stay under one hour. Automation, ranking control and personalization were non-negotiable must-haves, especially with future growth in mind.

After extensive testing, FactFinder came out on top. Since switching to this AI-powered solution, Popken has delivered more relevant results, seen stronger conversion rates and drastically reduced indexing times, from a targeted ten minutes to just two.

Product sorting also saw a major upgrade. In the past, results were always sorted by newest arrivals. Today, Popken uses a hybrid of self-learning AI and targeted business rules to adapt to local market demands, from promotions and stock levels to regional trends. But for the team, this was just the first step in a much broader product discovery strategy.

Challenge #2:
Delivering personalized fashion experiences at scale

Popken’s catalog is vast, with countless categories and style variations. At the same time, every customer has unique preferences. As Jason Ballard puts it, “How do we make sure every visitor finds the right product right away?”

It’s a simple question, but one with a complex answer. Relevance isn’t just about matching keywords or showing popular items. It’s about context. Two customers searching for “pants” might want completely different things: one a casual pair of jeans, the other something sleek for a special event.

Personalization is in Popken’s DNA. With FactFinder integrated across the group, the foundation was there. But what would it take to turn that vision into real impact, and how could the team prove its value, both for customers and the business?

Solution #2:
True personalization meets smart testing strategy

With the upgrade to FactFinder Next Generation, the team unlocked new strategic capabilities. Chief among them: Atlas AI, a machine learning model that interprets customer behavior in real time and learns from individual preferences.

This allows Popken to create personalized, 1:1 shopping experiences from the very first session, even with minimal data. For example, if a shopper browses mostly in the “Sports” category and later searches for “pants,” the AI will prioritize sports pants, not a random mix of jeans, business and casual styles. If the customer focus shifts, the sorting dynamically adapts.

Advanced analytics and A/B testing are also key. “The switch to Next Generation helped us zero in on our KPIs,” says Jason. “Channel-specific conversion rates in particular are a big focus for us.”

The team tested conversion rates with and without Atlas AI using extensive A/B tests, tailored to each market. The intuitive FactFinder backend made it easy to roll out changes globally or target individual shops.

Drawing from experience, Jason knows a one-size-fits-all approach won’t work: “One of the most common mistakes is taking the winning setup from one country and applying it everywhere. Every market and every customer group behaves differently.”

In some markets like Italy and Finland, personalization with Atlas AI is intentionally kept low. “Customers there aren’t as responsive,” Jason explains. “In places like Germany and Austria, we test more aggressively and see strong results.”

In Germany, for example, the personalization level (called “Atlas Impact”) was around 10 percent. In Finland, it was limited to 3 percent. “Once we find the sweet spot,” says Jason, “most of our A/B tests with Atlas AI deliver great results.”

Results

Integrating FactFinder marked a key turning point. The team finally had a scalable product discovery solution built to support global growth, complete with high performance and a user-friendly setup.

The rollout of Atlas AI and data-driven sorting quickly delivered results. In a six-day A/B test with 1.5 million sessions, Ulla Popken increased its conversion rate by 4.3%. “That was our first big win with Atlas AI,” says Jason, adding that the team relies heavily on accurate data: “We use tools that let us calculate statistical significance down to the decimal.”

And shoppers clearly recognize the added value; personalization is seen as a benefit, not manipulation. All testing insights flow directly into ongoing optimization and help improve performance across every international shop.

Internally, the way the team works has also evolved. Whether it’s ranking logic, personalization or clearance strategies, discussions now revolve around data, not gut feelings. “When I joined Popken, I brought in data-first approaches from other companies, such as ranking by sales, newness, conversion rate or availability,” Jason shares. “We saw immediate performance gains.”

Popken’s favorite features

Atlas AI
A/B-Testing
Multi-Channel-Management
Vector Search (Coming soon)

Atlas AI

Atlas AI takes sorting to the next level with real-time, 1:1 personalization. It analyzes every session, including clicks, filters, and scroll behavio, reordering product results based on immediate buying intent.

Behind the scenes, a neural model builds a digital map of the assortment. It recognizes connections between attributes and user behavior, continuously improving sorting in search and navigation. What used to require lengthy manual optimizations or large amounts of data is now solved automatically according to the current, individual buying intent, for a maximally relevant shopping experience and higher conversions.

LEARN MORE

A/B-Testing

Popken runs systematic A/B tests with FactFinder to unlock full eCommerce potential. Whether it’s search, navigation, personalization or campaign tweaks, every change can be tested directly in the shop.

“It’s super easy to set up,” Jason says. The result: reliable data instead of subjective assessments. This makes it immediately apparent which measures increase KPIs such as conversion and AOV the most. Successful variants can be adopted with a single click, for measurable sales growth without the need for internal debate.

Multi-Channel-Management

To efficiently manage search and product discovery in all markets and languages, the Popken team relies on a sophisticated multi-channel concept. There are two channel types in FactFinder: parent and child channels. Parent channel settings, such as synonyms, filters, or marketing campaigns, are automatically inherited by the associated child channels. This allows central optimizations to be rolled out across countries and ensures a consistent customer experience. At the same time, child channels can be customized as needed to flexibly implement regional requirements.

Vector Search (Coming soon)

Popken is already planning to implement FactFinder's new vector search with the aim of better understanding imprecise search queries or natural language inputs and minimizing zero-hit searches. “As soon as we can test vector search, we hope to eliminate this problem as much as possible,” says Jason.

Vector search captures the meaning, intent, and need behind every search query. On the results page, it reliably delivers relevant product results, even if there is no linguistic match in the data feed. The result: lower bounce rates, higher revenue and a more positive brand experience.

LEARN MORE

What's next

With more than 20 years of eCommerce innovation under its belt, Popken continues to raise the bar. And the focus remains the same: smart, data-driven product discovery that meets every customer’s needs.

“We’ve already picked the low-hanging fruit,” says Jason, convinced of Atlas AI’s long-term value. “AI-driven tools are essential now. They simplify our daily work and help us uncover patterns we’d never spot manually. If you have access to this kind of tech, use it.”

Personalization with Atlas AI works significantly better than traditional approaches. For us, AI-driven tools have become indispensable. They simplify daily work and uncover patterns we’d never spot on our own. If you’ve got access to this tech, you should absolutely use it.”

Jason Ballard

Onsite Manager

Learn how other leading retailers use FactFinder

Marvin Wagner,
Online Marketing & eCommerce Manager

“We’re definitely satisfied. With FactFinder, we have a partner that continuously evolves. The combination of smart technology and analytics is opening up entirely new possibilities for us.”

Mario Zanier,
CTO

"The enhancements extend beyond financials. The streamlined search and navigation have cut search times by 16%, and even more so on mobile, by 21%. Customers now browse 25% fewer pages to find what they’re seeking, with a 40% reduction on mobile devices."

Melanie Giebler,
Team Lead Product Management

"With FactFinder, we have found a partner that sets the same high standards for its services and products as us. They’re super cooperative and solution-oriented and will definitely be with us in the future.”