Reed.co.uk helps people find the jobs they love and want with Elasticsearch

For Reed.co.uk, the largest online job site in the UK, search technology is critical for the success of the business. With 11 million registered candidates applying for almost 100,000 roles every day, the speed and relevance of a results page can have a dramatic impact on careers and livelihoods. Search is also a priority for 30,000 recruiters seeking the best people for their business and who pay to list their vacancies on the platform.

To stay ahead in the highly competitive recruitment market, Reed is leading the way in the deployment of artificial intelligence, including generative AI, to transform its search capabilities. Krishna Chalasani, Software Development Manager for the search team at Reed, says, "Our two main objectives are to serve relevant results to customers and increase revenue through promoted jobs. Finding the sweet spot between these goals is our main challenge."

Elasticsearch plays a crucial role in this non-stop approach to innovation. When Reed first moved away from relational databases for search, it tested several alternatives. "When it came to performance and relevancy, the early results from evaluating Elasticsearch as a vector database were amazing," says Chalasani. Since then, Reed has taken advantage of frequent feature updates that deliver better results to users. "We simply update our Elasticsearch clusters to implement the latest releases. It's incredibly easy."

Chalasani emphasizes how Elasticsearch enables Reed to capture semantic relationships and meaning in a far more nuanced way than pure keyword matching. When a user begins their search, keywords are processed by a machine learning model to generate vector embeddings that capture search intent. Similarly, job descriptions are transformed into vector embeddings representing their requirements. Using vector similarity techniques, Elasticsearch matches user intent with job descriptions. Elasticsearch provides the ability to store embeddings in a vector database. The Elastic API has powerful search retrieval capabilities including filtering, aggregations, and role-based access controls at the document level.

If a job listing includes 'reports to director of engineering' this may not be relevant to the role itself. Vector embeddings and search tools in Elasticsearch enable Reed to make these subtle, but important distinctions and return more relevant results to the user.

Telemetry data captured by Elastic Search enables Reed to evaluate and compare the performance of job roles served to candidates. From information such as click-through rates, Reed can generate metrics that show how well a recommendation is performing. "As we uncover correlations for higher performing listings, we can use this knowledge to boost the relevance of less successful postings," says Chalasani.

More relevant searches also enable Reed to retain and acquire corporate clients who pay to promote job openings. "By adding vector matching to keywords in Elasticsearch, we can deliver more accurate results for hiring organizations. When they see a return on their promotional spend, they are more likely to continue their partnership and payments with Reed," says Chalasani.

When candidates search for a role, they can set an alert to receive updates for similar openings via email or push notifications. Elasticsearch is used for both search page results and alerts ensuring a connected user experience throughout the job search journey.

"Let’s say you get an alert for a new job on your smartphone, but you don’t have time to click the link straight away," says Chalasani. "When you return to your laptop a few minutes later you’ll see the same role on our website. Elasticsearch ensures that candidates can move between channels without interruption."

In the future, Chalasani also sees potential for generative AI to boost search relevancy further. "Vector matching is our priority, but in the next year we want to implement techniques such as learning to rank and generative AI that enable us to further refine search and deliver more personalized job results for candidates."

• See how Elasticsearch can work for you with a 14-day, no-obligation trial
• Learn how Dutch recruiter Randstad harnesses Elastic Security to protect tens of thousands of job candidates and customers from cyber attacks
• How Orca leverages Search AI to help users gain visibility, achieve compliance, and prioritize risks