Published on
June 11, 2024
.png)
Welcome to an exclusive interview with Julian Hensolt, the Founder and CEO of Dresslife. Julian's journey from working at Mercedes-Benz to pioneering AI-driven personalisation in fashion highlights his deep commitment to addressing the inefficiencies in the fashion supply chain. Through Dresslife, Julian is using advanced AI technology to transform the fashion retail landscape, improving both customer experience and sustainability. Read on as he discusses the inspiration behind Dresslife, the challenges of capturing shopper preferences, and his vision for the future of fashion retail.
Your journey from working at Mercedes-Benz to founding Dresslife is fascinating. What inspired your interest in leveraging artificial intelligence for personalization in the fashion industry, and how do you see this technology evolving in the broader context of e-commerce?
Julian: My academic and professional background played a significant role in inspiring my journey into leveraging AI for personalization in fashion. During my tenure at Mercedes-Benz, where I worked for over a decade and completed my PhD in supply chain management, I recognized several disadvantages in the fashion supply chain compared to other segments. Fashion is often produced in fixed batch sizes, leading to substantial overstocking issues. Additionally, statistics show that people wear 75% of their clothing fewer than four times in their life. The process of finding clothing is also more time-consuming than other products, resulting in missed revenue opportunities.
Fashion is a complex product segment, but I believed that insights from other supply chains could be transferred to the fashion sector. AI provides the tools necessary to achieve a significant leap forward. Thus, I founded Dresslife to offer a more convenient shopping experience that improves resource utilization. Personalization emerged as an essential lever for this change. By accurately predicting what customers will buy and keep, we can enhance revenue, reduce returns, and optimize product inventory.
Looking into the future, personalization is just the first step. We aim to extend our predictive capabilities across the value chain of fashion retailers. This could involve predicting demand or forecasting which designs a retailer should provide. Our ultimate goal is to develop an operating system for advanced fashion technologies that not only improves sustainability but also enhances profits, benefiting retailers facing tight margins.
Personalization in fashion involves understanding individual styles and fits. Can you share some key learnings or challenges you've encountered in developing AI algorithms that effectively capture and adapt to the ever-changing preferences of online shoppers?
Julian: Developing AI algorithms that capture and adapt to fluid online shopper preferences has been both challenging and enlightening. One of the most critical learnings has been distinguishing between a customer's style preferences and their immediate shopping intent. While these two elements may seem closely related, they serve very different roles in the personalization process. For instance, a customer might generally prefer blue suits, but this information is less relevant when they are currently seeking swimwear. This realization highlighted the necessity of distinguishing between enduring style preferences and momentary shopping intent. By fine-tuning our algorithms to accurately interpret and respond to a customer's immediate needs, we have significantly advanced the effectiveness of our AI, making our recommendations more relevant and responsive to the shopper's current journey.
Addressing fit is another complex and nuanced challenge. Fit encompasses the physical attributes of the human body, clothing design, and psychological elements of human behavior and perception. We place significant emphasis on the psychological aspect, understanding that fit is not just about physical measurements but also about meeting the customer's expectations.
The fashion industry is known for its dynamic nature and short product life cycles. How has your work in AI navigated the complexities of predicting trends and making accurate recommendations, especially considering the brief window products are often available online?
Julian: The transient nature of fashion and shorter product lifecycles present a significant challenge for applying AI in this industry. The rapid turnover of items, such as a t-shirt available for only a few weeks, makes prediction much harder. This contrasts with more stable products, like gaming consoles, where historical search trends can predict future interest with greater certainty, like the consistent demand for PlayStation during the Christmas season year after year.
To overcome this, we have developed an AI that significantly enhances product understanding. By analyzing every piece of data available—including images with and without models, garment measurements, product descriptions, and material details—we can make predictions for items even before they are sold. This approach not only counters the inherent product turnover challenge but also paves the way for a more intuitive and predictive online shopping experience, tailored to both current trends and individual preferences.
Reducing returns due to fitting issues is a significant achievement. Could you shed light on the key strategies or methodologies employed in your work that have successfully contributed to minimizing returns and enhancing the overall customer experience in online fashion retail?
Julian: Our primary goal is to elevate the gross profit per user after returns by balancing revenue enhancement and return reduction. We aim to optimize for profit by finding the combined optimum. One tactic we employ is the strategic sorting of our category feed. We prioritize items with the highest likelihood of being purchased while also considering the probability of returns. This method enhances revenue, mitigates return rates, and bolsters customer satisfaction.
To achieve this delicate balance, we analyze a variety of data points, including clothing dimensions, reasons for returns, and product images. This comprehensive approach ensures that we're not only driving financial gains but also delivering a shopping experience that aligns with customer preferences and expectations.
Beyond the immediate impact on returns and revenue, how do you envision the long-term effects of AI-based personalization on the fashion retail landscape? Are there broader implications for supply chain management, production forecasting, or other aspects that your work might influence in the future?
Julian: Beyond the immediate benefits of improved profits, the strategic application of AI promises to revolutionize broader aspects of the industry. From supply chain management to production forecasting, the insights generated by AI can enable a more agile and responsive fashion ecosystem. By facilitating accurate demand predictions and providing real-time feedback to designers, AI can help streamline production processes, minimize returns, and align product offerings more closely with consumer demands. This shift not only optimizes operational efficiencies but also supports the move towards a more sustainable fashion industry. In the long run, the adoption of AI in fashion retail could lead to more innovation, sustainability, and customer-centricity, reshaping the industry landscape in profound ways.
As we wrap up our interview with Julian, we want to say a big thank you for sharing your amazing journey and insights. Your work in using AI to personalize fashion is truly impressive. What you’re doing at Dresslife is making a real difference, and we appreciate your participation in Pixyle AI’s interview series. Thanks for being a game-changer in the fashion industry!
Pixyle AI generates e-commerce product data that enables brands, retailers and marketplaces to deliver exceptional product discovery experiences.
