Lokalee decreases search time by 85% and boosts sales by 37% with MongoDB Atlas Search

THEIR CHALLENGE

Enhancing Lokalee’s customers’ experiences with MongoDB Atlas Search

The travel and tourism industry is focused on creating meaningful and memorable experiences. In particular, hotels look to stand out from competitors by offering services that add value and deliver unique enhancements to visitors’ experiences. Some hotels have the in-house capacity to source and deliver these enhanced services. Those that don’t turn to Lokalee.

Lokalee sources and sells tickets for activities, tours, museums, and excursions in cities across the Middle East, Europe, and Asia. Unlike direct-to-consumer platforms, however, Lokalee white-labels its solution for use by hotels, enabling them to run an online platform under their own brand that generates income and can act as an online landing page.

“Some hotels are having problems financially because of businesses like Airbnb,” explained Mohamed Daher, Chief Technology Officer at Lokalee. “They want to find new sources to generate income, so the commissions they earn through Lokalee are very appealing to them.”

Lokalee launched in 2021 using MongoDB Atlas as its main database, along with aggregation pipelines and geospatial indexing to handle searches and filtering.

“We use a JavaScript stack, so MongoDB’s NoSQL approach is very easy to work with, especially with the Mongoose library,” said Daher. “We also realized that all these tickets in different cities and destinations would create a big database. That led to the decision to choose MongoDB Atlas.”

The platform worked well for over two years, but a change of curator—Lokalee’s main ticket supplier—saw the business’s inventory rise dramatically from 20,000 tickets in 77 cities to 350,000 tickets in 220 countries. This led Lokalee to rethink how its platform would operate.

“We decided to make a new version of our product, and it was a six-month effort,” Daher added. “We redesigned and rebranded, changed some of our server structure, and overhauled both our back end and front end. The only thing we didn’t change was the database.”

The increase in business inventory, however, meant that database updates and search processes using the aggregated pipeline approach suddenly became more complex, and tasks were more time-consuming.

“We had our tickets in one collection, our venues in another, cities in another, and so on,” explained Daher. “To make the expanded inventory work, we had to run three to five searches on each collection, and it was taking a lot of time.”

The result was a risk that buyers would lose patience and drift away to another supplier. Lokalee needed a quick, agile, and effective solution.

THE SOLUTION

MongoDB Atlas Search to the rescue

Daher tried an open-source solution that appeared to meet Lokalee’s needs, but quickly ran into a key issue.

“We had to push all the data in our MongoDB database to their database to run the search. I didn’t want to have to manage and operate separate databases, so we were stuck. It was Elliott Haddad our MongoDB account manager who came up with the answer.”

The proposed solution was MongoDB Atlas Search. It offered a much faster and more responsive search functionality than the previous aggregation pipeline approach, and was also a much better match for the new inventory’s tagging system.

“Viator has a mesh tagging system that allows tickets to be allocated simultaneously to multiple subcategories, which is a great help to the search and filtering functions,” said Daher. “Without MongoDB Atlas Search, however, really making the most of this would have taken too much time.

With the help of MongoDB Professional Services, Lokalee was able to complete the implementation and migration within just four months.

“Aditya Dua, our Professional Services contact was very helpful and understood the project very quickly,” Daher noted. “He gave us specific recommendations on facet-based searches and dealing with tokens in Atlas Search. In one call we went over all our implementation, he gave us some recommendations, we applied them, and ‘Boom!’ Everything changed and we had a system we could live with.”

THE RESULT

Key Improvements to search times, user experience, and sales

With MongoDB Atlas Search, Lokalee’s relaunched platform has a vastly expanded choice of experiences—and sales opportunities—with a slick search and filtering system that makes tracking down the perfect choice quick and easy. Searches that once took 20 to 25 seconds, and would even sometimes time-out after 30 seconds, are now completed in three or four seconds.

“The searches are all dynamic, too. If a city doesn’t have any tickets of a specific type available, they don’t show up as options,” Daher said.

The new platform immediately delivered a 37% increase in sales. In addition, the developer experience was also transformed, with Daher and his lean team now able to remove multiple tasks from their workloads.

“We don’t need to handle multiple pieces of middleware anymore, or check their validations, indexing, or performance,” he added. “For the back-end team, things are so much easier and faster. Issues that might once have taken Maroun and George two days to resolve can now be fixed in five minutes.”

However, the great strides forward are just a step along the way in Lokalee’s evolution. Daher now plans to make use of AI to develop a comprehensive trip planner tool that will use large language models (LLMs) to build complete itineraries, including ticket sales, based on individuals’ criteria for each trip.

“We’ll convert from the normal JSON (JavaScript Object Notation) database to a MongoDB vector database with embeddings which we can then integrate with local LLMs,” Daher concluded. “This is definitely an exercise that I’m going to run with MongoDB and its AI track, and MongoDB Atlas Search is also bringing real value to that solution.”