Iterating the search engine

Over the last few months, we've been working hard on our Planning Application Aggregator, adding new features like corporate subscriptions, as I mentioned in my previous posts. Today, I'm excited to share some updates on how we've supercharged our search engine, boosting both speed and quality to improve user experience significantly. As our user base grew, both on the individual and corporate fronts, it became increasingly evident that our search feature was turning into a bottleneck. Users were facing slow search speeds and, occasionally, less-than-accurate results. Given that search is a core functionality of our aggregator, it was crucial to tackle this issue head-on.

This was in essence a lot of testing real life queries with several experimental db queries. The disjointed, and sometimes downright bizarre, nature of Planning Application data meant that the queries needed to be quite general but also account for countless edge cases.

After rigorous testing, we rolled out the new search engine to all our users. We monitored the performance closely for the first few weeks, prepared to rollback if things went south. Fortunately, that hasn't been necessary so far.