This is the story of how Hunkemöller’s merchandising team worked together with an AI model, delivered by ADC Data & AI Consulting, to achieve better discounting outcomes and deliver stronger sell-through at higher margins.
Joost van Rens, Chief Operating Officer at Hunkemöller, explains what the numbers showed and how the merchandising team changed its approach after adopting artificial intelligence.
Results of the transformation
Hunkemöller’s merchandising team carried out the transformation with ADC, guided by a benchmark that ADC sets for the retail industry over a three-month period:
- Revenue up 8 per cent.
- Margin up 2 percentage points in three months.
The pilot ran from mid-November to the end of January—about two and a half months. In that time, Hunkemöller not only reached the benchmark, but exceeded it—faster than expected, on both revenue and margin.
Today, AI-driven markdown runs across every Hunkemöller channel—online and in stores, in every country where they operate. To anchor the change, ten key people were retrained on the new pricing principles.
Hear the Hunkemöller case live
RetailTrends Conference, 13 November.
Why Hunkemöller chose to transform markdown with AI
A deep review was the trigger. Markdown across the company was increasing and showed persistent negative trends:
- Full-price sell-through percentages were falling.
- Average discounts—and the share of high-discount tiers—were rising.
Hunkemöller operates in many countries and across both offline and online channels. The complexity of predicting the right discount for each market and channel was a Herculean task, and as a result, its average discount sat above historical norms.
The message was clear: too much was being given away. Stronger demand forecasting was needed to avoid clearing stock at excessively low prices.
A gut feeling became a structured challenge—one that called for a systematic, AI-driven approach to markdown.
How markdown worked before AI
Before the transformation, pricing and markdown were fully manual. The team relied on experience. Decisions were made step by step as each sale unfolded. The full season was not planned in one go. Meanwhile, as scale and the need for tighter control grew, managing at SKU or fine product-group level in Excel alone became too difficult.
So Hunkemöller turned to ADC to implement and test an AI approach to discount management.
A/B testing to determine the impact of working with AI
To see whether the new approach would bring measurable benefits in a short period, the team started with A/B testing of SPARO—Strategic Pricing and Revenue Optimisation. SPARO is a fully customisable decision-support system that blends machine learning with business logic, delivering pricing recommendations in structured “waves” rather than one-off changes. It is designed to balance speed and margin protection—moving stock efficiently without discounting too deeply.
SPARO was used throughout the entire sales period—starting with Black Friday and running all the way through the end-of-season sales—across both channels (online and in stores) and in every country.
To get an objective view, they set up test and control groups:
- Test: ADC’s discount recommendations plus the merchandising team’s judgement.
- Control: the traditional approach by the merchandising team alone.
They did not just compare discount depth; they also tracked margin and volume, as it is easy to cut deeper and lose units. On all three metrics, the results from combining the merchandising team’s expertise with SPARO were significantly better.
What changed in the merchandising team’s workflow
With SPARO, Hunkemöller added a new step to its pricing cadence:
- ADC generates recommended discount levels.
- The merchandising team reviews them and gives feedback.
- Where context requires it, the team applies targeted overwrites; otherwise, recommendations are accepted as they are.
Early on, reviews sometimes took longer and created friction. Today, the process is smooth—reviews and overwrites happen on time and without issues.
Because SPARO has already proven its advantage, control-group checks have been retired. The approach is now standard practice under a three-year strategic partnership with ADC.
What has changed in markdown philosophy
The planning logic has shifted. Previously, each sales stage was handled independently. Now the entire period is optimised as a single plan. The new flow is deliberate: the highest discounts are reserved for a small portion of end-of-line stock—capturing more sales earlier and lifting overall margins.
Control is more granular: discounts are tailored to specific items and situations, keeping reductions precise rather than broad.
Samantha Swain, Chief Merchandise Officer
Our collaboration with ADC was successful. We established a strong relationship quickly and worked together effectively to deliver results. The use of SPARO for the end-of-season sale was particularly productive.
Once the model learned our parameters, it made a real difference. It pushed us to go deeper on the right items and move faster. It also removed a lot of administrative work, allowing us to focus more on analysis.
Event-driven promotions such as Black Friday are more challenging because they limit the model’s flexibility. This was due to internal complexities related to retail layouts and workload, requiring closer collaboration to achieve results.
This is a great tool. If we were asked where we think the solution works best, we would say pure e-commerce, simply because the tool then has fewer constraints and can operate all year round. However, we find it a great support for true clearance activity.
Joost van Rens, Chief Operating Officer
AI for markdown works best when there’s scale in assortment and data. You don’t need to be international—models run per country, with smaller markets grouped where appropriate.
What matters is the foundation:
- Two or more years of clean transaction, price, and inventory history.
- Hundreds of products per season or category.
- Enough price variation to estimate elasticity.
- Clear stock signals and channel flags (online/stores).
Adoption is typically easier in e-commerce—no shelf labels to reprint—so levels can be very precise. In stores, tiered levels keep the offer clear for shoppers.
This is not about replacing people. It is about a stronger partnership: the merchandising team working with SPARO. ADC introduced their model to us; we added Hunkemöller’s constraints and business rules, tested it, saw the uplift, and adopted it as our standard.
If you are considering it, start with a fit check with ADC: a quick data diagnostic and back-test to confirm signal strength—then a focused pilot during a sales period.
See Hunkemöller and ADC on stage at the RetailTrends Conference on 13 November
Full AI markdown case with live Q&A. Invitation only. Apply for a pass.
What stage is your organisation in on its data-driven journey?
Discover your data maturity stage. Take our Data Maturity Assessment to find out and gain valuable insights into your organisation’s data practices.
