Update to Google Suggest

There's been quite a bit of comment in the last few days about Google Suggest, particularly how it's used in Google Chrome. Google Suggest is actually built into a number of different products including Google Search, Google Toolbar, browsers like Google Chrome and Firefox, and the Google Search application on the iPhone.

But what is Google Suggest? It's a feature designed to make searching both faster and easier. Whether you are typing into the search box on Google Toolbar or Google.com, or the Omnibox in Google Chrome, Google Suggest guesses what you're typing and offers suggestions in real time. So for example, if you type "bass," Google Suggest might offer a list of refinements that include "bass fishing" or "bass guitar." Similarly, if you type in only part of a word, like "prog," Google Suggest might offer you refinements such as "programming," "programming languages," "progesterone," or "progressive." Again, these suggestions allow you to enter your searches faster, but also in a way that will more likely get you the results you're looking for.

That's what occurs on the surface of Google Suggest. Here's what happens under the hood. To provide its recommendations Google Suggest needs to know what you've already typed, so these partial queries are sent to Google. For 98% of these requests, we don't log any data at all and simply return the suggestions. For the remaining 2% of cases (which we select randomly), we do log data, like IP addresses, in order to monitor and improve the service.

However, given the concerns that have been raised about Google storing this information -- and its limited potential use -- we decided that we will anonymize it within about 24 hours (basically, as soon as we practically can) in the 2% of Google Suggest requests we use. This will take a little time to implement, but we expect it to be in place before the end of the month.

All data retention is a balance between user privacy and trust on the one hand, and security and innovation on the other. In the case of Google Suggest we decided it's possible to provide a great service while anonymizing data almost immediately. But in other cases - such as our core web search - storing data like IP addresses for a time is crucial to make improvements to search quality, improve security, fight fraud and reduce spam.

You can read more about our privacy policies on the Google Privacy Center, check out the Google Search Privacy: Plain and Simple video for more information about how our server logs work.

Update October 5, 2010: We launched Google Instant last month, and we now store Google Instant's partial query data for up to two weeks in unanonymized form, at which time we will delete 100 percent of it. These data retention changes apply only to queries made when Google Instant is active.

Posted by Urs Hölzle, Senior Vice President, Operations