I Love Search Suggest

The first time I saw Google Lab’s suggest function, I spent about two hours just playing around with it. Granted, my idea of fun differs from most people’s, but I really think this is an important innovation in search practices. It’s a great example of a technology that solves a problem so effectively that it’s bound to become ubiquitous.

By now, you’ve probably seen this: it’s a list that drops down from a search box as you type, showing query strings similar to the one you’re typing. In usability testing of search engines that don’t have suggest functions, I’ve seen many users struggling to formulate just the right combination of words that will produce matches for the information need locked up in their heads. Search suggest makes that task easier.

As users become accustomed to selecting suggestions from this list, you can expect that their queries will converge on fewer and fewer phrasings, which will in turn make optimization efforts more effective since people will use phrasings that have already been optimized more often.

I’m not sure who thought of it first, but you’re starting to see suggest everywhere in slightly differing flavors:

  • Google suggest lists queries that other people have tried, in descending order of popularity. That approach makes good sense: if many users are searching for something, it’s more likely that you’re searching for it too. As a result, the suggest function often seems to anticipate what you’re trying to find.
  • Ask.com’s implementation is very similar to Google’s, but lacks the annotation listing the number of results you’ll get if you submit that query. I think that’s unfortunate, because its presence on Google Suggest helps users evaluate the specificity of the search before submitting it.
  • Yahoo’s version matches against words in the middle of the search string, instead of just those that are at the beginning. This seems like a very good idea, because it removes the somewhat unfair burden on the user to type the right words first. But it’s also a more complicated algorithm, and I’d be curious to learn about its reliability.
  • Google Finance’s version may belong in a different class of suggest functions altogether. While it looks similar to other implementations, it’s only providing suggestions for known items in its database. When you submit a suggested string, you don’t get a list of search results; you get a single page for that item. This opens an important question about what the user’s typical expectations are for suggest functions. In this case, the function is promising “these are things for which there is a known match,” but in the cases above it’s just saying “these are searches that other people have tried”. I wonder if that dichotomy causes problems for users.
  • Apple has, as it always does, gone all out and included short descriptions, images, and groupings in its drop-down. Apple mixes known items with search result lists, but moves you strongly toward picking queries with one best answer. It’s an ambitious approach, and I think it’s fantastically successful.

Depending upon the implementation selected, a suggest function can also be relatively easy to build. Given the high ratio of benefit to complexity, I would expect to see suggestion features become more or less standard over the next few years. Let’s say that by mid-2010, your site will start to look behind the times if it doesn’t have one.


No comments yet

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: