Case Studies
Measured Results From Randomised Split Tests
Every number on this page comes from A/B tests with real shoppers. No cherry-picked cohorts. No projections. Just randomised splits and measured outcomes.
Case Studies
Every number on this page comes from A/B tests with real shoppers. No cherry-picked cohorts. No projections. Just randomised splits and measured outcomes.
Germany's largest second-hand fashion platform. 2,000+ brands, 1M+ products. Every item is unique - once it's gone, it's gone.
The paradox of choice. With over a million unique items and no duplicates, shoppers couldn't find what they wanted through traditional navigation or keyword search.
Wordless Search deployed in days with no manual tagging required. Miros AI reads product images and shopper behaviour to surface relevant items without words.
50/50 randomised split test
"A strategic step that transforms our business."
Selim Yurguc, Senior Business Development
140-year-old brand. 1,000+ SKUs across puzzles, games, and creative products. A household name in Europe and North America.
Puzzles are chosen by image, not by keyword. Traditional search buried products because shoppers couldn't describe what they were looking for in words.
Wordless Search powered by Miros image analysis. Shoppers browse visually and the AI learns their preferences in real time.
50/50 randomised split test. Now expanding to US market.
"Miros provides a way of presenting our wide range in a method consumers control."
Mark Sutherland, Head of Global DTC eCommerce
Singapore's largest marketplace. Luxury bags category with millions of active users and high-value transactions.
Needed better product recommendations for luxury buyers. Existing algorithms weren't capturing the visual preferences that drive luxury purchase decisions.
Head-to-head test against Carousell's existing recommendation algorithm. Miros deployed alongside for direct comparison on the same user base.
1M-user randomised split. Rolled out across entire luxury range.
"Miros outperformed our existing algorithm by 31+%."
Ashutosh Shukla, Director of Product Management
Furniture retailer. High-value, infrequent purchases where the challenge isn't the first sale - it's getting buyers to come back.
Converting first-time buyers into repeat customers. Furniture shoppers typically buy once and disappear to competitors for their next purchase.
Intelligent Feed activated from the very first visit. Miros builds a preference profile in real time, making subsequent visits immediately relevant.
24K buyers tracked over 3 months
"I forgot other stores existed."
ON24 shopper
US second-hand fashion marketplace. 200,000+ unique items at any given time. Every piece is one-of-a-kind.
Years of internal attempts at improving product discovery had drained budget without results. Traditional approaches couldn't handle the scale and uniqueness of inventory.
Wordless Search narrowed 200K items down to dozens of relevant products per shopper, based on real-time behavioural signals rather than keywords.
140K visitors, 50/50 randomised split
"Miros has higher conversion than our site search."
Gray King, CEO
Operating in Spain, France, and Germany. 10-year-old second-hand fashion platform. Had already tried and failed with 3 previous vendors.
Paradox of choice combined with over-reliance on discounting to drive sales. Products existed in the catalog but shoppers couldn't find them without price cuts.
Wordless Search enabled shoppers to find relevant items in approximately 60 seconds, reducing dependence on discounting as a discovery mechanism.
153K visitors, 50/50 randomised split, one month
"Not because the price is too high, but because they're not being found."
Lourdes Ferrer, CEO
Aggregate impact
Randomised A/B split tests, 140K-2M sessions per test
Working demo on your catalog within days. We'll run a head-to-head test against your current discovery stack.