Google Wants You to Do Side-by-Side Comparisons in Search
Google has launched several new additions to its enterprise version of Labs today. First and foremost, they've created a new tool that will let businesses compare any enterprise search engine with Google Search Appliance or Google Site Search. The software will then let you administer a anonymized and randomized test that users can vote on, letting you determine the quality of search results from different options.
The other announcements today are a new Google Search Appliance Connector for the Salesforce CRM, and the full launch of Google Apps Script for all Premier and Education Edition customers. Apps Script has been in a pilot since May but is now up to snuff for production environments.
Side-by-Side Search Comparison
Google's new download, which they're literally calling Side-by-Side, is designed to test both the front end of various searches and the quality of the results themselves. A very basic explanation of how it works would be that administrators download the tool, assign some search policies to compare, and then users vote on the results.
The tool is aimed at those using either Google Site Search or Google Search Appliance — we're not talking about Google versus Bing — letting them compare those options either with a different implementation of the GSA or another enterprise search.
The Trust Issue
Usually when a company offers a "compare our product" resource of some kind, you pretty much know who's going to come out looking better. But this isn't a simple comparative matrix of features: it's a customer-controlled testing tool.
Google clearly knows that fixing the results would be shooting themselves in the foot. Two other aspects of Google's new tool make me think there are other reasons to trust it:
It's anonymous. Adding Google's logo and UI to any kind of search results automatically increases user satisfaction in consumer tests. But in the Side-by-Side tool, this dominance in branding can be masked by giving both sets of results the Google look. That means if you want to compare quality of results and ignore the UI, then you can.
You can properly randomize results. Google has done a good job of letting you avoid common problems that occur in side-by-side tests. For example, you can randomize which side each set of results appears on, thus reducing user habituation to placement.
All in all, this new tool should be a fairly easy way for IT to tinker with their search architecture in the enterprise. By packaging a system for administrators to rapidly test results with users, the new tool should be a useful addition to the toolset for enterprise search operations.
