CUSTOMER STORYUsing AI to eliminate search dead ends for one of the world's top five retailers.

How one of the world’s top five retailers used AI-powered semantic vector search to dramatically reduce zero-results queries and boost revenue.

The customer journey roadblock

Online shoppers expect to interact with technology using natural language, typing or speaking queries the way they’d ask questions of a friend. When a customer arrives at a website and searches for something, they expect results that match their intent—not just their exact words.

But what happens when a shopper searches for a specific brand you don’t carry? All too often, they hit the dreaded “no results found” message—a critical dead end in the customer journey. The reality is, you might have exactly what they need, just under a different brand or description. If your search engine isn’t smart enough to understand the problem behind the query, valuable connections—and sales—are lost.

This was precisely the challenge facing one of the world’s top five retailers. Their inventory consisted of a wide selection of different products but limited varieties of each product type. Traditional keyword-matching search resulted in countless queries returning zero results, and as the saying goes: if customers can’t find it, they can’t buy it.

The limitations of manual curation

The retailer had previously migrated their product catalog search from Endeca to Apache Solr, but the implementation essentially replicated the old system rather than improving its capabilities. Without machine learning, search results required manual curation through an ever-growing set of rules.

For the retailer’s modest search team, maintaining this complex web of rules became exhaustingly labor-intensive. For customers, the experience often led to poor result relevancy or no results at all, causing frustration and high bounce rates. And for the business, this translated to high operational costs and substantial revenue losses.

Rather than diverting essential employee time to become search experts internally, the retailer sought specialized partners who could implement and manage a highly-scalable search solution that would resolve zero-results queries, improve relevancy, and reduce the manual curation burden. With its foundation in Solr, track record of improving retail product discovery, and AI-powered platform, Lucidworks was the perfect fit.

Diagnosing the search failure points

Lucidworks identified several query types that weren’t producing results on the retailer’s site:

• Brands or items not carried in their inventory
• Items in high demand whose stock fluctuated frequently
• In-store only items not appearing in online searches
• Vocabulary mismatches including:
  • Long-tail queries (“weatherproof fleece-lined pants”)
  • Thematic queries (“warm jackets”)
  • Symptom queries (“dry cough, acid reflux”)
  • Misspellings
  • Exact queries like product model numbers

The Never Null solution

To address these challenges, Lucidworks deployed semantic vector search with a proprietary deep-learning encoder called Never Null. Unlike traditional keyword matching, Never Null learns from customer behavior to associate queries with products that have a similar purpose—even when the exact terms don’t match.

The system uses behavior signals to train search models, continually tuning and improving results. This means the search box can yield results based on semantic meaning rather than simple keyword matching, dramatically reducing “no results” scenarios.

For the retailer’s search team, this meant freedom from the repetitive task of analyzing results and manually creating rules. The system’s advanced machine learning automatically retrieved relevant results for challenging queries, allowing the team to focus on more strategic initiatives.

By broadening the fields searched and understanding semantic relationships, the retailer’s product assortment in search results expanded significantly, driving increased engagement, conversions, and average order value (AOV).

Putting AI to the ultimate test

The true test came during Cyber Five—the five-day period from Thanksgiving Thursday through Cyber Monday, retail’s most intense online shopping period. As the typical morning average of 250 queries per second ramped up to 1,250 QPS, the system maintained performance with 99% of requests processed in just 100-300 milliseconds.

The importance of search became even more evident as engagement jumped from a pre-event daily average of 15% up to 31% over the Cyber Five. Never Null’s out-of-the-box models performed impressively even for new shoppers with no signal history, making them feel like familiar customers.

Over the five-day period, the system handled an astonishing 680 million search requests without a single outage. The results were remarkable:

• Zero-results queries reduced by 91% compared to the previous year
• Search-influenced orders increased by 30%
• Search-influenced order value grew by 28%

Beyond the immediate wins

As Never Null continues to learn from customer behavior signals, the retailer now has a solution that evolves with their customers’ needs. The system’s ability to understand customer intent rather than just keywords means shoppers are discovering more relevant products—whether they ask for them by exact name or not.

For a retailer with broad selection but limited varieties in each category, this semantic understanding has transformed the customer experience from one of frustration to one of discovery. No longer hitting dead ends, customers are finding what they need, filling their carts, and completing purchases at significantly higher rates.

The success demonstrates a fundamental truth in today’s digital retail landscape: search isn’t just a utility—it’s a critical revenue driver that requires sophisticated AI to truly understand and serve customer intent. By investing in semantic search technology, this top-five retailer has not only eliminated the dreaded “no results” message but has created a foundation for continued growth and customer satisfaction.

---

By deploying semantic vector search with Never Null technology, one of the world’s top five retailers transformed their digital discovery experience during their busiest shopping period, processing 680 million searches without a single outage and reducing zero-results queries by 91%—proving that in modern retail, the difference between a bounce and a purchase often comes down to understanding what customers really want, not just what they say.