An EU apparel retailer replaced JoliSearch with BRADsearch. A/B test across 3,200 visitors and 12,670 searches - zero-result rate dropped from 20% to 6%, CTR up 54%, conversion up 30%.
This is an EU-based online apparel retailer specialising in a niche fashion catalogue - clothing, footwear, accessories, and home décor - shipping to customers across Europe, the USA, Asia, Australia, and Africa. Their buyers are not casual browsers. They know what they want: a specific style, a specific brand, a specific garment in a specific aesthetic.
That purchase intent is an asset - if the search bar can meet it. With a niche catalogue, a worldwide customer base, and highly specific product vocabulary, search quality is not a peripheral concern. It determines whether a buyer who arrives ready to purchase finds something to buy.
The retailer's customers are international, and many type in English as a second or third language. Spelling variations that feel natural to a human reader - a misspelt style name, a product typed with a typo, a term spelled wrong - returned nothing under JoliSearch. One in every five searches ended with zero results. That is a buyer who arrived ready to purchase and left empty-handed.
This niche apparel catalogue has its own language: highly specific subcategory styles, aesthetic labels, brand-specific terminology, material descriptors. JoliSearch matched keywords. It could not understand that closely related style queries pointed at the same product type, or that a use-driven term should surface the right footwear results regardless of how the catalogue tagged them.
With a genuinely global customer base, GDPR compliance was not optional. The retailer needed search analytics that did not depend on cookies - which meant JoliSearch's tracking approach was structurally unsuitable regardless of its other limitations.
BRADsearch's typo tolerance operates at a level that handles real-world international spelling variation, not just single-character edits. Misspelt style names, mistyped product terms, and garbled brand words all resolve to correct, in-catalogue products. This single capability was the primary driver of the zero-result rate improvement - from 20% to 6%, with further improvement continuing as the analytics reveal new edge cases to tune.
Product type, style, material, brand, and descriptive tags are all indexed and searchable. Customers can find products using the language they naturally use for this niche apparel catalogue - not the language the catalogue happens to use internally. The search engine bridges that gap rather than requiring customers to reverse-engineer product naming conventions.
GDPR-aligned analytics without cookie dependency. The analytics layer immediately began revealing which brands and clothing types generate the most search demand - and which products customers are actively searching for that the retailer does not yet carry. That second insight is direct catalogue intelligence: demand without supply, visible and actionable for the first time.
Zero-result rate dropped from 20% to 6% - a 70% reduction. That means 14 out of every 15 searches now return relevant results, compared to 4 out of 5 before. And the rate continues to trend down as analytics inform ongoing improvements.
For a store where purchase intent is high and product vocabulary is niche, eliminating failed searches translates directly and proportionally into revenue. The 12,670 searches analysed in this A/B test represent a statistically robust sample across a diverse international buyer base. Details are shared anonymously at the retailer's request.
← Back to all case studiesJoin merchants across Europe who use BRADsearch to turn every search into a sale.
✓ No credit card required · ✓ Free onboarding setup · ✓ Cancel any time