Why Redfin Switched from Microsoft Virtual Earth to Google Maps: A Performance Analysis

Redfin transitioned its mapping service from Microsoft Virtual Earth to Google Maps primarily to achieve significantly faster page load times and improved user experience. Internal performance tests revealed that while Google's initial script loaded faster, a custom coding solution ultimately made Google Maps the superior choice for displaying hundreds of property listings simultaneously. This decision was driven by quantifiable performance gains that are critical for a responsive real estate search platform.

Why did Redfin initially choose Microsoft Virtual Earth over Google Maps?

The original selection was based on a specific technical requirement: the speed of adding numerous markers to the map. Early testing consistently showed that Microsoft Virtual Earth was up to four times faster at adding a high volume of items—a crucial capability when displaying up to 500 homes on a single map view. Although Google Maps loaded its base script and map tiles more quickly, its performance lagged during this intensive operation. A key factor was a "bulk add" feature in Virtual Earth that Redfin had specifically requested, which addressed this core need at the time.

What changed to make Google Maps a viable option?

Contract renegotiations with Microsoft prompted a re-evaluation. A critical breakthrough came when a developer created a custom GOverlay for the Google Maps API. This solution replicated the bulk-add functionality, effectively eliminating Google's previous performance disadvantage. This innovation led to a prototype development effort to assess the feasibility of a full platform migration.

How did the technical migration from Virtual Earth to Google Maps proceed?

The transition was smoother than anticipated because Redfin's use of map APIs was relatively basic. The platform primarily needed to center the map, draw property markers and neighborhood polygons, and respond to user pan-and-zoom events. Both APIs handled these fundamental tasks well. Notably, the zoom level numbering scheme was identical, preventing issues with saved search URLs. The most complex challenges were unrelated to the map itself, involving data parsing and user interface management.

From a developer's perspective, the Google Maps API offered several quality-of-life improvements:

• Better Event Handling: Distinct events for programmatic moves versus user drags simplified code logic.
• Fewer Timeout Issues: The API design avoided patterns that caused delays in the previous platform.
• Lighter Operations: Actions like resizing the map container were less resource-intensive.
• Cleaner Code Practices: Google Maps was more respectful of global JavaScript and CSS namespaces, preventing potential conflicts.

What were the final performance results after switching to Google Maps?

The performance gains were substantial and justified the migration. Comprehensive testing revealed:

• The map page's onload event fired 2.5 seconds faster due to Google's efficient content delivery network (CDN).
• In a worst-case scenario test adding 500 houses, Internet Explorer 6 saw a 10-second speed improvement. Faster browsers like Firefox 3 still showed a 3-second gain.

These improvements translate directly to a snappier, more efficient user experience, which is a top priority for online real estate searches. The final transition from a working prototype to a production-ready feature took two weeks of rigorous testing and refinement, ensuring reliability for hundreds of thousands of users.

For anyone evaluating technology platforms, the key takeaway is to regularly re-assess options based on core performance metrics. What was once a technical limitation can be overcome with new solutions, and even small speed improvements can have a major impact on user satisfaction.