Has there been any study into how quickly autocomplete suggestions need to show up for a good user experience?


3 Answers 3


Giving immediate results with first character is the way to go, according to me.

If the search history is saved for the particular user (Like in google if your are logged in, or on your Mac/Windows, where you're the only user) then this is a valuable thing to have. Most of the time in chrome/spotlight I just type one letter and hit return for my frequent websites/applications. Like F+return will get me to facebook in less than a second. I also feel that it sort of makes my experience more personal, I know this is my system and it knows my usual requests and is expediting the process.

There is also the notion of the impatient user. As the systems get more powerful and the bandwidth increases, people are getting less tolerant of wait-time. No one likes watching the "Loading..." prompt. Why I bring this up is, if you delay the search by a couple characters, it might lead the user to believe that the system is slow/not good enough to guess things right off the bat.

The downsides is, you need a fast enough system to make it a fluid experience (else you will end up with a flickering search results list).

Edit: Just remembered Nielsen's time scale explanation for UX. http://www.nngroup.com/articles/powers-of-10-time-scales-in-ux/

It resonates with what I wrote above.

Instant results:

0.1 second is the response time limit if you want users to feel like their actions are directly causing something to happen on the screen . For example, if you click on an expandable menu and see the expanded version in less than 0.1 seconds, then it feels as if you made the menu open up.

To create the illusion of direct manipulation , a user interface must be faster than 0.1 second.

1 sec results:

When the computer takes more than 0.1 second but less than 1 second to respond to your input, it feels like the computer is causing the result to appear. Although users notice the short delay, they stay focused on their current train of thought during the one-second interval.

This means that during 1-second response times, users retain the feeling of being in control of the interaction even though they notice that it's a 2-way interaction (between them and the computer). By contrast, with 0.1 second response times, users simply feel like they're doing something themselves.


We presume that

  • an autocomplete is essentially a search operation, and
  • that users understand that they are searching.

Though I don't have a paper to cite, I have seen that around a 1000-1600ms window is range before which people begin to purse their lips or otherwise get 'the stare'

If this is a multi-use field that does not absolutely present itself as a 'search box' then you can probably have it delay a little longer.

Generally, even with apps with slower UX allowances, a search field involves a more hurried state of mind.

  • 3
    1 second would be pretty long for simpler autocompeltes. As I note in this answer many big-name apps have effectively instant autocomplete
    – Zelda
    Commented Mar 28, 2013 at 13:40
  • Wow, I wouldn't think that >1000 ms response time would be acceptable for autocomplete, especially with Google's being so stinking fast. Commented Mar 28, 2013 at 13:40
  • @BenBrocka All politeness intended to you and OP, but if this was a big-brand app I don't think the question would be asked as it is here. Users' expectations are set by many things, including product & brand image. I wouldn't even expect a McDonald's site to have lightning autocomplete; 'internet utilities' is not what they 'sell' Commented Mar 28, 2013 at 13:48
  • That said, I think the 'lightning' autocomplete experience is a reasonable ideal. I've even heard of people engineering a page to include a quick-load set of some 100 or so terms, so that the autocomplete fills in with something instantaneously, and then updates with more as the AJAX results come in. Commented Mar 28, 2013 at 13:52

I am not aware of any studies on this subject. A search of journal archives (in particular, ACM SIGCHI) did not yield anything useful - a handful of studies that looked at autocomplete mechanisms as part of their studies, but none that answered this question specifically.

HCI is still a young field, and published answers to questions about particular widgets and their design are generally exceptions, rather than the norm.

Not the answer you're looking for? Browse other questions tagged or ask your own question.