How Tamaris uses AI for Predictive Pricing: a conversation with 7Learnings at NRF Retail Big Show

Paris - At the NRF Retail Big Show, FashionUnited spoke with Eiko van Hettinga, Co-Founder of Berlin-based tech company 7Learnings, which provides AI-powered predictive pricing software for retailers. He explained why pricing is one of the biggest levers for profitability, how predictive pricing works in practice, and how brands like Tamaris are seeing results.

Why Predictive Pricing?

“Pricing is the biggest lever for profitability. Many companies think of cost cutting when they talk about profit, but the truth is that price movements have a much larger impact. That’s why analysts like Gartner call price optimization one of the most attractive AI use cases in retail: it has the highest business impact and is one of the most feasible.”

“For retailers wondering where to start their AI journey, price optimization should be at the top of the list. With predictive pricing, you use data to forecast the impact of prices on your KPIs - like revenue, margin and sell-through - and then optimize accordingly. That’s exactly what we demonstrate together with Tamaris.”

Can you tell us more about Tamaris?

“At Tamaris (part of the Wortmann Group), the challenge was clear: they were expanding their online business into 26 countries, with huge complexity in setting prices across markets and channels, a lot of manual work, and the need to optimize along the entire product lifecycle.”

“Together we ran a five-month proof of concept, and the results were striking: profitability increased, their average discount rate went down by 5 percent and the manual time spent on price optimization was cut in half. Today, Tamaris runs this AI-driven setup across all their markets, steering prices and margins automatically.”

So it’s not only about discounting? Is it also about flexible pricing?

“Exactly. It’s not just about putting a big red sale sign in the window. You can also lower prices strategically and still maintain your margins. In fashion retail, we do everything for fashion reasons, so we think more broadly than just discounts.”

What makes online pricing more complex?

“Online, you have additional layers like vouchers and coupons. The danger is that if you start stacking too many promotions, you can easily lose control of your profitability. That’s why we feed this kind of data into the system as well to predict how much of these promotions will be used and what impact they will have on profit.”

“In fashion retail we also forecast return rates together with prices. Across the full assortment, our forecasts reach more than 90 percent accuracy, that’s for a two-week horizon. We believe the right approach is a combination of highly accurate short-term forecasts and longer-term planning.”

Why is that?

“You could say, why not use a 40-week forecast to make every decision? The problem is that such long-term forecasts are very inaccurate you simply don’t know what will happen that far ahead. That’s the big challenge in fashion.”

“We use long-term forecasts to set boundaries, not to dictate every decision. For example, the algorithm might calculate the price that optimizes long-term profit, and then allow us to move within a 20 percent range around that point. Within this range we can make short-term decisions, like pushing sales faster, but the system also prevents us from going so far that we hurt long-term profitability. Technically, we believe that’s the best way to solve it — and practitioners in the field confirm this approach.”

Are you also involved in collection development, deciding which styles and quantities?

“Not really. Once the collection is out there, we can help with initial pricing, but that part usually has more of a human touch. If you have a dress entering the market for the first time, we can look at its attributes and compare it to similar items to suggest a price. But if you believe it’s a standout piece, then human judgment comes in, because the machine won’t see that. Over time, as the product goes through its lifecycle, the system learns more and more from transactions and attributes to improve its pricing decisions.”

And do you have a fashion background?

“Our CEO, Felix Hoffmann, has spent his whole career in pricing. He first worked for consultancies like A.T. Kearney, and later was responsible for the pricing algorithm at Zalando in Berlin. At some point, he realized you can’t just work with Excel forever — you need something more technical. That’s how the idea for 7Learnings was born. Today, we are an independent company.”

7Learnings has also worked with retailers such as Tom Tailor and Mister Spex, helping them implement predictive pricing. The startup was founded in Berlin in 2019 by Felix Hoffman, Eiko van Hettinga and Martin Nowak.