Impact

Design went from invisible to embedded in every revenue-critical decision. Here is what that produced: measured, tested, and validated at every level of the product.

The context

HSE is one of Germany's largest home shopping platforms: ~€800M in annual revenue, reaching ~46 million households across Germany, Austria and Switzerland via TV, with 17,000 SKUs across fashion, beauty, jewellery, and home.

When I joined in 2020, design was a 2-person shared-service function. Designers served multiple product managers reactively, called in after decisions were made, asked to produce screens, not shape strategy. Design had no ownership of outcomes, no seat at discovery, and no way to make its impact visible.

I joined to change that. What follows is what changed, and what it produced.

The case studies show the individual contributions. Each one has a clear problem, a specific decision, and a measured result.

Impact at a glance

Team & organisation

  • 2 → 9 Team scaled over 6 years
  • 0 → 1 First dedicated User Researcher in company history
  • Stage 2 → Stage 4 UX maturity over 6 years (Nielsen Norman Group framework)
  • Replaced a shared-service model with 8 designers owning outcomes inside cross-functional product squads, one per squad, involved from discovery through release.
  • Secured and onboarded HSE's first dedicated User Researcher, integrating research into product decisions from the start rather than at the end.
  • Design maturity moved from reactive, delivery-only execution to a structured, embedded practice with dedicated research, governance, and measurable accountability. NN/g maturity model →

Design system & delivery quality

  • ~15% Faster delivery across squads after adoption
  • ~25% Fewer UI regressions on high-risk journeys after adoption
  • HSE's first design system: shared tokens, reusable components, accessibility standards, and contribution governance, across web, iOS, and Android. Delivery speed improvement measured after adoption across 8 product squads.
  • Fewer regressions followed system adoption: shared components carry correct states, behaviours, and accessibility standards by default. Teams that migrated stopped rebuilding the same elements from scratch and stopped inheriting inconsistent behaviour in the process.
  • Read the full case study →

Customer satisfaction · CSAT

  • ~2pp CSAT improvement per year · consistent 6-year trend
  • ~76% Starting baseline in 2020
  • CSAT was not tracked when I joined. I built the case for it with a straightforward argument: without a longitudinal satisfaction signal, there is no way to know whether what we ship improves the experience or erodes it. In e-commerce, dissatisfied customers do not just churn; they tell others. CSAT gave design a metric that spoke the language of the business. Leadership approved it.
  • Measurement method: Hotjar post-interaction surveys, triggered immediately after successful purchase and payment completion. Timing is everything: CSAT surveyed days later measures memory, not experience.
  • The data was operational, not decorative. Low satisfaction scores on the checkout journey surfaced payment friction that drop-off data alone could not explain. That signal directly informed the decision to expand payment options, fixing a problem before it became a revenue issue.
  • Honest caveat: CSAT captures how users felt at a specific moment but cannot isolate design's contribution from pricing, logistics, or customer service. We triangulated it with usability testing and task completion rates to avoid acting on noise.

Search · highest-intent journey

Search · Elasticsearch rebuild · 2024

  • +4% Click-through rate on search results
  • +8% Customer satisfaction with search
  • −42% Engineering maintenance time
  • Search is HSE's highest-intent journey. Customers use it four distinct ways: by product number from TV, by brand, by category, and by open-ended query. Defining success criteria separately for each intent type was the decision that unlocked the improvement.
  • The −42% engineering maintenance reduction was a deliberate design target, not a side effect. Easier-to-maintain search means faster iteration, which compounds over time.
  • Read the full case study →

Product page · returns and content

Product page · 2023 to 2024

  • −19.5% Ring return rate reduction
  • ~€100k/year Agency costs eliminated · AI copywriting tool
  • A size-finding UX improvement on ring product pages reduced return rate by 19.5%, without changing the product. Returns are a direct cost: logistics, restocking, lost revenue.
  • The AI copywriting tool generates descriptions and SEO texts for ~17,000 SKUs. Research validated AI copy matched human copy on trust and purchase intent before it shipped. ~€100k in annual agency costs eliminated.
  • Read the AI tool case study →

Checkout · conversion at the highest-stakes moment

Guest checkout · A/B tested · Amplitude · 2024

  • +6.6% Conversion rate uplift · A/B validated
  • Users had to be logged in to buy. That single requirement was costing conversions at the worst possible moment in the funnel. Three versions were designed and tested. The winning version moved account creation entirely out of the pre-purchase flow.
  • Conversion improvements at checkout scale directly into revenue; every percentage point matters at HSE's order volume.
  • Read the full case study →

Social commerce · HELLO App · 0→1

HELLO App · 2020 to 2024 · Approved by C-level

  • 60M+ Viewers in Germany · 300+ creators · 5,000+ live shows
  • HELLO was a 0→1 social commerce app, approved by C-level and one of Europe's largest live commerce platforms. Built to validate a new product category and reach a 25 to 35 audience that was not engaging with traditional teleshopping.
  • The product worked. The go-to-market did not: a planned marketing investment failed to materialise, and the assortment was not adapted for the target audience. Both constraints were outside design's scope.
  • The strongest product patterns (entertainment-first feed, creator identity as a trust signal, in-stream commerce layer) were subsequently integrated into HSE's core platforms.
  • Read the full case study →

Shake It! · second-screen feature · Silver Award

Shake It! · 2020 · Silver Award · Eyes & Ears

  • ~25% Conversion uplift in a quarter
  • 4.8/5 App rating · FY2020
  • One gesture from live TV to the product page: open the app, shake the phone, land on the PDP. Removed the search step entirely for the highest-intent moment in the HSE customer journey.
  • Singled out in HSE's official 2020 annual financial results press release. Won the Silver Award at the International Eyes & Ears Awards 2020.
  • Read the full case study →

What the numbers don't show

Every metric above required the same precondition: design being involved early enough to influence the decision, not just the screen.

That precondition didn't exist when I arrived. It had to be built, and building it was harder than any individual product problem I worked on. It required changing how product managers thought about design, how engineers worked with designers, how leadership measured design contribution, and how designers understood their own accountability.

The three concrete changes that made everything else possible:

The metrics above are the result of six years of that work. They are not a coincidence.

Takeaway

Measurable design impact doesn't happen because designers work harder. It happens because the organisation lets design work earlier, on the right problems, with the right accountability, and the right evidence. That is what I build.