Beni unifies fashion resale to boost sales by 40%

THEIR CHALLENGE

Making second-hand shopping simple

Beni’s mission is to make second-hand fashion as easy to buy as new. That means handling the complex logistics and technology behind resale every day.

“Beni is a second-hand search engine that helps users find resale alternatives while shopping online,” said Beni’s CTO and co-founder, Celine Lightfoot. “We started as a browser extension that surfaces second-hand listings during the shopping flow, and we also have an app and a website where you can shop our catalog directly.”

Unlike traditional fashion operators that sell thousands of identical units, the second-hand market consists of “one-of-one” inventory scattered across platforms such as eBay, Poshmark, and Depop. To bring this vast resource of items to a wider target market, Beni must manage a massive, fragmented, and easily searchable catalog of over 300 million listings that refreshes by more than a million items daily.

That challenge alone is substantial, but managing a fractured data architecture made the task even more complex. Beni launched using Milvus, a niche vector database, where it stored vector embeddings before setting up a separate MongoDB instance for catalog attributes.

“We rely on vector search. We’ll use a combination of a reverse image search, or an image plus text,” said Lightfoot. “We then generate an embedding of the product information, add weights to it, and apply that to a catalog using vector search.”

But maintaining two distinct data stores created synchronization issues within its extract, transform, load (ETL) pipeline; simply adding a new attribute to Milvus—such as 'style type'—involved rebuilding the entire database from scratch, a process that took days and hampered Beni’s end-user functionality.

“That’s really tricky for a company that’s trying to innovate quickly,” explained Lightfoot. “We’re constantly evolving and we need to generate new attributes in our data ingestion pipeline so it’s easier for us to run more interesting and engaging queries.”

Scalability and cost efficiency are also key considerations for Beni. Its business model relies on commissions, so its infrastructure costs need to be both scalable and sustainable.

“Our technology needs to be very affordable for the unit economics to make sense,” added Lightfoot. “Cost is a very big piece for us, and we saw that standardizing into one database would help us optimize that.”

OUR SOLUTION

A powerful, unified, and simplified solution

By unifying its database environment with MongoDB Atlas, Beni was able to consolidate its vector embeddings, product catalog, and user engagement data into a single database, effectively replacing the old, fractured dual-system architecture. The move immediately streamlined Beni’s operations by simplifying its ETL pipeline.

“It is significantly easier to keep one data store live versus having two separate data stores and needing to keep them both live through the ETL,” said Lightfoot. “It’s also significantly more cost-effective than trying to maintain two separate catalogs.”

To handle the sheer volume of data generated by the second-hand market, MongoDB also provided robust sharding capabilities that met Beni's critical scaling issues. Its 300 million-plus listings exceeded the RAM limits of even the largest search nodes, so sharding allows Beni to distribute this data across smaller, more cost-effective instances. Lightfoot describes the approach as essentially “organizing inventory into smaller boxes” rather than renting one massive warehouse.

“The sharding process was complicated, but we were able to work through it seamlessly with support from the MongoDB Flex Consulting team. Our Consulting Engineer, Jay Chakra, was amazing and incredibly helpful,” said Lightfoot. “The support enabled us to get to a point where we can scale our catalog endlessly.”

Unifying on MongoDB also unlocked powerful hybrid search capabilities. By storing interaction data alongside vectors, Beni can now execute complex queries that rank items by key criteria such as relevance and popularity. 

“With MongoDB's test, vector, and hybrid search capabilities, we’re able to pull products from our catalog that are both semantically relevant and also highly popular based on other user engagement data,” said Lightfoot. “That enables us to deliver a more personalized, more relevant search experience.”

OUTCOME

Reduced costs, improved clickthroughs, more sales

The result for Beni is a transformed working environment that aligns with its core objectives of innovation and flexibility while also dealing with high and rapidly changing data volumes.

MongoDB accelerates innovation, allowing Beni to dynamically add new fields—such as 'style' or 'boldness'—to enable rapid iterations of its data model. Consolidating vector embeddings and product catalog into a single unified data store has also streamlined Beni’s engineering operations and eliminated the need to run two separate systems, resulting in a 20% reduction in monthly infrastructure costs.

“We’ve reduced duplication, simplified synchronization, and cut compute and storage overheads,” said Lightfoot. “With MongoDB we can scale our catalog efficiently without additional operational complexity.”

MongoDB’s sharding capabilities have resolved critical scaling limitations, and the migration has also unlocked powerful hybrid search capabilities that are directly improving key business performance. By storing user engagement data alongside vector embeddings, Beni now executes complex queries that rank items by both semantic relevance and popularity, effectively turning its choice of infrastructure into a direct driver of sales.

“MongoDB enables rich and contextually relevant results for shoppers,” said Lightfoot. “We’ve seen up to a 43% improvement in our image search clickthrough rate (CTR), and a 30% improvement in our query research clickthrough rate. Those increases in engagement are translating to an increase of around 30-40% in total sales, with a steady order value.”

In the past two years Beni has diverted over $5 million in gross merchandise value to the resale market, and it’s a success that’s set to continue. Beni is now looking to migrate from MongoDB 8.0 to MongoDB 8.1 to simplify its data pipelines even further, use AI to drive personalization, and make its hybrid search function even more powerful as the company plans its expansion.

“With MongoDB, we’re continually focused on expanding our product catalog, increasing filterability and the number of dimensions we have with all our data stores,” noted Lightfoot. “We think we’re the best positioned to win in this space, and we’re excited to be on the journey with MongoDB.”