Today we rolled out an update to HTR's hotels database and hotel lookup on the site where we migrated from Google Places as the core search to Tripadvisor.

TL;DR

• At the core of HTR is our hotels database
• Since inception, we have used Google Places as our hotel lookup to find hotels
• PROBLEM #1. The issue with Google Places as our core lookup is that its not specialized for hotels which meant that the database was able to have hotels that weren't even hotels (ie. users could select their city or any other type of location in Google) and Google often doesn't have the best hotel specific content
• PROBLEM #2. In order to launch HTR's upcoming AI Tech Stack Advisor tool we needed a stronger foundation of a hotel lookup that could accurately on-the-fly build a model of a hotel with key attributes like location, size, ADR, chain scale, amenities and hotel type while ensuring that the only results in the hotel lookup field were real and valid hotels
• New Hotel Lookup Component. As such, we rebuilt our entire hotels db and hotel lookup throughout the site with a new component built on a foundation of Tripadvisor's API with custom logic and mappings for things like keywords, attributes and guest type to further enrich and segment the database and ensure accurate hotel lookup and model building in real time on-the-fly
• Historical Data Migration. A major undertaking in this project was to run the old hotels database of 100k+ hotels through the new hotel lookup component to verify, validate, enrich and de-dupe hotels which impacts historical data in the database 

How will this impact my company?

Migrating historical data is a massive undertaking and will be rolled out over the next several weeks gradually updating ~5,000-10,000 hotels per day.  During this time you will likely notice some changes to the flags on reviews on your profile as existing/legacy hotels in the database get mapped to Tripadvisor hotels and then enriched via the new hotel lookup.  If you have any concerns or see any data that you believe to be inaccurate please let us know.