An In-Depth Look at Search Auto-Suggestions

Search auto-suggestion features are a recently hot search trend. Basically, there exist three approaches to search suggestions:

• Completion (to make users’ lives easier by saving keystrokes)
• Suggestions (to remind users of their previous related searches, fix their spelling mistakes and make suggestoins for popular/related results)
• Instant results (to let users find something without even searching by letting them jump to the search result right from the search box)

Let’s see how search auto-suggest features have been implemented throughout the web.

Google’s Multi-Purpose Auto-Suggest (“Search Suggestions”)

Google, the number one trendsetter in search, has successfully combined the latter two approaches. Its auto-suggest feature has been hugely enhanced over the past years. In many cases, it removes any need to search at all by showing suggestions based on the user’s most likely intended query:

These are most frequently used to help the user search more efficiently by:

• Suggesting popular search patterns
• Reminding the user of his previous searches
• Fixing incorrect spelling

In fact, it is no longer necessary to search in order to see spelling suggestions; these are now available as you type. What is even more impressive – these search suggestions are based on context:

Google will also help you jump automatically to the site or page you need, bypassing the actual search results page. This feature is enabled for “navigational” queries where the user’s intent is clearly to navigate to a web page:

The algorithm behind pulling search suggestions has been widely discussed – mainly because the search auto-suggest does influence the searchers’ behavior and hence the way they get to your site using Google search. In fact, Google auto suggest may actually hurt your brand image:

It may also completely direct an uncertain train-of-thought or alter the direction of conversations:

There is no official statement as to what exactly influences Google’s auto-suggest results but it’s been observed that the two main factors are:

1. Global: Popular search (what exactly users type). The good news (actually that’s no news at all and has been around for a couple of years) is that Google suggests really fresh terms – which makes it very useful for observing brand (hot) context as well as researching trending topics.
2. You (Personalized): Your search history (mainly, what you most recently searched for) as well as your location. These results are usually displayed over much more globally popular results.

2. Facebook “Socially Enhanced” Auto-Suggest

Facebook is another huge player that quickly adopted the auto-suggest feature. Facebook’s algorithm is a bit different – it is primordially based on your connections:

you will instantaneously see results not only of the people, events, groups and Pages you’re connected with but also the connections of your friends and globally relevant results.

Hence, we can call these “socially enhanced” search suggestions.

The anatomy of Facebook’s search suggestions is as follows:

• Results are all mixed (friends, Pages, answers and apps will all appear as you type)
• Connections are consistently displayed on top  (above Pages and everything else)
• Your connections define the order: for example, when it comes to Pages, you first see those you like and then pages your friends like (as well as exactly which friend likes it)

3. Examples of Creative Search Suggestions

Let’s take a look at a few more examples from around the web and see how search auto-suggest features may enhance user experience.

1. Usability law: If a user gets lost on a website, they will leave. Users don’t want to learn  how to use the site; they need to be able to instantly understand it.

Solution: Search suggestions enhanced with navigation. This way the user knows where he will be directed to without before clicking “Search” – and he will instantly get an idea of the site architecture, topics and categories he might be interested in.

FanSnap implements the category search which makes the option more usable and the site structure easier to understand:

2. Usability law: Users almost always scan – they rarely read carefully online.

Solution: Make your search suggestions easier to scan by adding visual enhancement: display logos and product previews. Icons make search intuitive, they draw attentions and make it easier for the user to understand where he is going to land.

My Bank Tracker has a very engaging search suggestions feature. It actually might draw your attention even before you hit “Search”. The only trick? All they do is display a tiny bank logo next to each suggestion that matches the characters you are typing:

3. Usability law: Users don’t know what to search for:

You are lucky if the user comes to your site with a clear idea of what he is looking for – in this case he will most probably find it. But what he is not so sure? What if he is, in fact, clueless?

Solution: Auto-suggestion functionality that gives users clues of where to browse further and suggests results from various levels in the site structure.

Last.fm search suggests artists as well as tags that start with the same characters, thus giving the users more ideas of  where they can browse further.

Have you ever spotted any creative usage of search auto-suggest feature? Do you think search auto-suggestions help improve usability or do they hurt it?  We would love to hear your thoughts!