How Toom increased user learning with GrowthBook

Executive Summary

Toom, one of Germany’s largest DIY retailers, reached a point where their experimentation program could no longer support the learning velocity they needed. Optimizely’s traffic-based pricing held them to testing on only 5% of visitors, which slowed significance, limited the types of experiments they could run, and made it difficult to challenge assumptions or understand nuanced user behavior.

After moving to GrowthBook, toom unlocked full access to 100% of website traffic and accelerated every part of their experimentation lifecycle. Their partner Creatistas, played a major role in this shift. Creatistas are experts in GrowthBook who have built sophisticated AI-driven workflows and automated reporting that help toom plan, launch, and interpret experiments quickly and consistently. Their collaboration has been essential in helping toom reduce operational friction and focus on what matters most: learning.

Within the first four months, toom realized their opportunity was far bigger than expected. GrowthBook creates a path for individual product teams to experiment directly, use feature flags, and eventually compare entirely different shopping cart experiences rather than simple UI tweaks. They are building toward a culture where learning is continuous, assumptions are challenged regularly, and every meaningful feature is tested.

‍

Why Toom Needed a Change

1. Traffic constraints reduced learning

Optimizely’s pricing model forced toom to cap tests at only 5% of traffic. That led to slow significance and prevented the team from exploring smaller but meaningful behavioral effects.

2. Slow iteration on UI testing

A complex workflow for UI tests created delays, added friction to launching new ideas, and prevented product teams from owning experimentation.

3. Heavy analytics burden

Optimizely required a second tracking implementation in parallel with GA4, and internal teams had to maintain custom analytics logic to analyze experiments. This added operational cost and discouraged more ambitious testing.

4. A desire for deeper learning

Toom wanted to drive a broader transformation. The goal is not simply to run more tests but to build a culture where teams learn quickly, update assumptions often, and deeply understand user behavior.

‍

"We knew we were not learning fast enough. Limiting tests to 5% of traffic meant we could not challenge assumptions or understand users at the level we wanted."
— Lukas Jungmann, Lead Customer Value Creation, toom  

‍

Why Toom Chose GrowthBook

Unlimited traffic for experimentation

Toom can now run experiments on 100% of website visitors. This opens up smaller effect size testing and dramatically shortens time to significance.

Native integration with GA and BigQuery

GrowthBook fit perfectly with toom’s mature analytics setup. They can build custom metrics, analyze retroactively, and avoid maintaining duplicate tracking setups.

Self-hosting ready

Toom prefers to self-host tools like Jira for security reasons, and GrowthBook’s architecture fits that requirement perfectly.

Strong partnership with Creatistas

Creatistas brought extensive GrowthBook expertise, enabling:

• Automated experiment analysis using AI agents
• Standardized Jira workflows for documentation and reporting
• Reliable measurement using existing GA signals
• Faster turnaround from hypothesis to decision
  
  

‍

What Toom Has Achieved So Far

A hypothesis driven experimentation process

Toom places careful effort into designing experiments. Every test begins with a clear hypothesis and a well defined metric strategy. This discipline ensures that when they see a winner, they know exactly why. Outcomes are currently: 55% winners, 11% losers, and 33% neutral. Toom is driving more winners and losers through tighter hypothesis and success definition.

Big Win: 95% growth in loyalty card signups

A desktop notification promoting toom’s loyalty card produced a 95% uplift in signups, generating 600 to 800 incremental enrollments in just a few weeks and no negative impact on revenue or conversion rates.

This overturned a long-held internal fear that notifications would annoy users.

Google Ads landing page improvements

Visitors who landed on the wrong product page from Google Ads tended to bounce. Toom tested a simplified layout with more space and a similar products slider above the fold.

The result was a major improvement in session continuation and new confidence to adapt this layout for standard product pages.

Surprising learnings

1. Countdown timers and “3 hours left” messaging produced almost no impact
2. Pop-up notifications had significant positive impact despite internal concerns

This reinforced the importance of evidence-based iteration rather than assumptions.

Operational Excellence

Creatistas acts as an extension of the toom UX team. Their automated workflows use GrowthBook data, N8N, and seven specialized AI agents. This setup generates analysis, metrics, checks, and recommendations in two minutes instead of three to four hours.

‍

"GrowthBook gives us freedom, and Creatistas helps us turn that freedom into a disciplined, scalable experimentation program."
— Lukas Jungmann, Lead Customer Value Creation, toom  

‍

A Bigger Opportunity Ahead

Only a few months into using GrowthBook, toom realized that the future is much larger than simply testing UI changes. Toom’s next phase includes:

• Running experiments inside the toom mobile app
• Using feature flags for controlled rollouts
• Testing full basket and checkout flows with URL splits
• Equipping product teams to run tests independently

The long term vision is simple and ambitious:
Every feature gets tested. Every product team is empowered to learn from data.