29/07/2026 at 2:00-3:00 PM (CEST)
This page includes everything you need to revisit the session, share insights with your team, or pass it along to colleagues and partners. You’ll find the full recording, key materials, and an easy way to get in touch. If you'd like to continue the conversation, our team is just a click away.
<aside>
Maximilian Lohse (ABxpert) and Frederick Mieke ran a joint webinar in which they analyzed two product detail pages live (Mustang and Bodylab) and shared concrete optimization hypotheses drawn from around 3,000 experiments. The focus was on topics like trust elements in the ATF area, videos to reduce return rates, sales-psychology-driven price communication, and the optimal buy box structure. In the second part, Frederick showed how these hypotheses can be set up and iterated as A/B tests directly in Claude via natural language using the Varify MCP.
ATF area (Above the Fold)
Trust elements like shop ratings belong directly in the header, complemented by the top 2 store benefits (static, not as a slider). Free returns should be prominently visible, not hidden in the footer.
Videos & return rate reduction (especially Fashion)
Integrate a video play button in the ATF area, primarily addressing fit. Tom Tailor case: AI-generated product images and videos delivered significantly positive results for more than 20 SKUs. A fit hint like "model is 1.90m, wears XL" helps reduce returns.
Color selection in the ATF
Customers constantly scroll up and down to change colors. Color selection belongs directly at the top of the buy box.
Social proof
Customer count mentions ("daily protein shake for over 2 million customers") placed above the product name. For Fashion, it is often better to work with shop ratings instead of product ratings (too few reviews at SKU level).
Price communication (correct from a sales psychology perspective)
Order: old price, new price, savings. Discounts in percent for larger amounts ("-18%"), sales prices without decimals ("79€" instead of "79.00€"). The Shopify default often gets this wrong.
Buy box structure (works across the board)
Price, delivery time, add-to-cart, payment icons (sorted by popularity: PayPal, Klarna, credit card), benefits.
Free shipping threshold
Place it 10 to 15% above the current AOV, ideally tested against profit.
Product attributes & quantity selection (Bodylab case)
Instead of a slider for quantities: show all variants directly, highlight the bestseller. Flavor selection visually with images instead of a dropdown, so users feel like exploring more varieties.
Creative idea: play back voice messages from influencers via UTM parameters ("Listen to why Lisa loves Bodylab").
Evaluation based on several criteria: visual difference, technical effort, content maintenance, position on the page (ATF vs. footer), expected win probability, quality of the underlying data. This produces a PRIO score per hypothesis.
Freddy showed live how an experiment can be created in Claude using natural language:
The full webinar recording is available on Youtube
https://www.youtube.com/watch?v=FqNjNd_tX7A
<aside>
🗓️ Book a 1:1 conversation with a member of our team: Click here
</aside>
<aside>
<aside>
<aside>
</aside>
</aside>
</aside>
<aside> 🟢
</aside>