E-Commerce Case Study
Pixum
How Pixum achieved a +16% conversion uplift through AI-powered optimization and personalization within 3 weeks
How Pixum achieved a +16% conversion uplift through AI-powered optimization and personalization within 3 weeks
Pixum is one of the leading online photo services in Germany and Europe. In addition to excellent customer service and a simple, inspiring shopping experience, Pixum delights its customers with high-quality branded products such as the Pixum Photo Book, Pixum Wall Art, and Pixum Photo Calendar.
For many years, the company has embraced data-driven development and fostered a culture of innovation. With over 200 A/B tests per year and a strong analytics focus, Pixum ensures the success of new features and changes, firmly embedding experimentation into the company’s DNA.
This mindset leads to well-founded decisions, faster insights, and continuous improvement of the website and products. With an openness to new approaches, Pixum embraced pagent.ai’s AI automation for copy and content optimization, with the goal of discovering what real value comprehensive content adjustments and personalization can deliver.
Pixum is among the leading providers of personalized photo products. Despite a strong brand and mature online shop, one central question remained: How do we convince users to start configuring their products when visiting the website and take the next step toward purchase?
After over 10 years of continuous optimization and verification, achieving significantly positive test results has become increasingly difficult. Accordingly, there was a natural skepticism: Can text changes alone really make a big difference?
Björn Prickartz - Head of Digital Analytics and Optimization at Pixum“While working with pagent, we quickly realized that we were adopting a new AI-first workflow where we co-create with AI and significantly increase our testing velocity. The consistently strong results reinforced our decision to use pagent for additional projects.”
With pagent, Pixum was able to expand and significantly accelerate its classic A/B testing setup. Instead of lengthy lead times for new content variations, relevant experiments are now launched within just a few days. pagent supports not only ideation but accompanies the entire process: from generating different variations to stakeholder alignment to rollout.
The self-learning optimization engine from pagent continuously analyzes user behavior and reactions to new variations. This enables pagent to deliver increasingly better optimization suggestions, continuously refine variations, and learn over time what works best for the customer.
Through this process, we achieved a high success rate for variations with positive impact. At the same time, a dynamic workflow emerged where AI and team collaborate, combining speed, creativity, and quality.
The results speak for themselves: Within just a few weeks, Pixum achieved three significantly positive test results with pagent. A direct impact on key KPIs such as click-through rate and revenue was observed.
| Hypothesis | Number of Text Changes | Conversion Rate Uplift |
|---|---|---|
| B: High-Quality Customization + Joyful Celebration + Creative Freedom | 34 | + 16.21% |
| C: Versatile Display Options + Creative Empowerment + Preserving Precious Memories | 33 | + 13.42% |
| D: High-Quality Customization + Friendly Guidance + Creative Freedom | 8 | + 5.05% |
The return on investment comes not only from revenue increases but also from sustainable learning effects: Each variation provides additional data points that improve the customer experience in the long term. For Pixum, the experiments with pagent represent a solution approach to achieve more significant tests again. The generation of variations that are initially detached from internal knowledge and constraints offers a fresh “outside perspective” and appears to be promising.
Björn Prickartz - Head of Digital Analytics and Optimization at Pixum“With pagent, we can establish an AI-powered system that enhances our experimentation process with fast, self-learning variations. This gives us the potential to implement incremental improvements that we wouldn’t have considered as hypotheses before.”
Pixum sees pagent as a strategic evolution of its optimization setup. The next step: Scaling context-aware and personalized variations further, based on channel, language, device, or user behavior.
Classic A/B testing evolves into a continuous test-and-learn system that enables personalized customer journeys across channels. pagent serves as a catalyst: From initial copy optimization to a flexible, AI-powered framework that secures long-term competitive advantages for Pixum.
Björn Prickartz - Head of Digital Analytics and Optimization at Pixum“The self-learning system promises to further minimize our work in the review process over time. We’re excited to benefit even more from time savings and automated testing in the future.”