AI in Fashion: How Hunkemöller uses AI for customer insight, price optimisation and store clustering

Artificial intelligence (AI) is rapidly changing the fashion industry. While some companies are waiting, others are taking bold steps forward.

In this edition of the AI in Fashion series, FashionUnited speaks with Gordon Smit, the technology director of Dutch lingerie company Hunkemöller.

1. How do you view AI and what does Hunkemöller use it for?

AI has become indispensable in modern organisations, especially in the fashion industry. Companies not yet working with AI are falling hopelessly behind. To remain competitive, AI must be embedded and continuously developed.

Our data team has grown from three to 12 people in a year and a half. We use AI throughout our entire value chain, from product development and design to sales and analysis.

2. Do you have any specific examples?

We are currently experimenting with 3D design in the design phase. By viewing products entirely digitally and in 360 degrees, we can drastically reduce the number of physical samples from Asia. We aim for one instead of four or five per design/style. This saves time and costs.

AI also helps us with image classification. Lingerie photos sometimes show a lot of skin, so Google can mark them as ‘adult content’, which negatively affects our discoverability. With AI, we can predict in advance which photos are likely to be rejected and which can be safely posted online.

Another important application is price elasticity. Take Black Friday, for example. Where we used to start marking down items sometime in November based on intuition, we now do it in a completely data-driven way. Machine learning models determine exactly when a product should or should not be marked down and by how much. This demonstrably yields better margins.

We also use AI for customer feedback. Together with Google, we developed a tool that automatically translates hundreds of thousands of reviews and measures sentiment. This helped us discover where the biggest customer frustrations were, allowing us to address them immediately.

Additionally, we are working on store clustering, where AI identifies which stores serve similar customer profiles. By grouping stores based on data, the product range per cluster can be much better tailored. These analyses sometimes require processing billions of records, a task that was manually impossible.

3. What has this AI journey yielded so far?

Hunkemöller has undergone a major data transformation in recent years. We had over 25 different data sources, which were consolidated into a single central database three to four years ago. We were sitting on a data goldmine but couldn't access it yet. Bringing all those sources together was a huge task, but now we are reaping the benefits. It has provided us with new information, such as patterns in shopping behaviour through store clustering.

The next step is to truly activate all these new insights, just as we did with customer feedback.

4. What lessons have you learned and what are the challenges?

The most important lesson is that your master data must be in order. If your data is incorrect, it remains a case of ‘shit in, shit out’. For price elasticity and store clustering, for example, we had to significantly tweak our data. To give you an idea, laying a solid foundation took us two years of blood, sweat and tears.

Another major challenge with AI lies in its adoption. Using AI within a large organisation is very different from how we use it privately. Asking everyday AI like ChatGPT to create a travel itinerary is simple; using it professionally is another matter entirely. For example, how do you ensure that 6,500 employees can write good prompts?

We are now developing training and guidelines to make employees more AI-proficient. We are also building a central AI strategy so that teams do not all work with different tools. This coordination is crucial, something many companies will likely recognise.

5. What’s next for Hunkemöller in terms of AI?

I just read a report stating that 90 percent of companies are already working with AI, but 67 percent of them are still in pilot mode. That is quite recognisable. In terms of insights, Hunkemöller is advanced, but in other areas, we are still in the discovery phase.

One of the areas we are just beginning to explore is creative AI. Although physical shoots remain essential for creating magic, emotion and atmosphere, AI can support and transform them in the future. It can expand creative possibilities or improve efficiency, for example, by reducing travel.

Additionally, we want to use AI to optimise our marketing mix and better understand the returns on our campaigns.

6. Where do you see the biggest opportunities for AI in fashion?

The biggest opportunities lie in the creative domain. Consider trend analysis: what should you develop; what designs are emerging; which direction is the market moving in? You can have AI create mood boards or convert patterns into a 3D design. This technology already exists but is still barely used at scale in fashion.

European players like Zara and Loavies and Chinese giants like Shein and Temu have very short lead times from design to delivery, often just a few weeks or days. We cannot match that pace. The design and production of lingerie are done entirely in-house and are more complex than making a T-shirt or a jumper. Nevertheless, our time-to-market can and must be shorter, and I am convinced that AI will play a key role in this.

7. Any final advice?

Last year, I said that companies should implement AI step-by-step: start small, run pilots and then scale up slowly. My thinking on that has completely changed. AI has given time a new dimension. A few years ago, ‘the past’ meant five, six or seven years ago. Now, when I talk about ‘the past’ in terms of AI, I am talking about two or three months ago. Developments are moving so fast that small steps no longer work.

For companies currently in the experimentation and exploration phase, ensure you have support within the organisation. Employees need to understand that AI is not taking over their jobs but freeing up time for them to do their work better. Especially in retail, where it is always ‘busy, busy, busy’, AI tools can deliver enormous efficiency gains.

For companies that have yet to start with AI, my tip is: data, in bold, underlined and with an exclamation mark!

AI tools were used to transcribe this interview and as a writing aid.

This article was translated to English using an AI tool.

FashionUnited uses AI language tools to speed up translating (news) articles and proofread the translations to improve the end result. This saves our human journalists time they can spend doing research and writing original articles. Articles translated with the help of AI are checked and edited by a human desk editor prior to going online. If you have questions or comments about this process email us at info@fashionunited.com