I could add the affordance help-text next to the field but on any page of the application there may be several auto-complete fields for different types of entities, so the text would become tedious and consume too much space.

So I’m considering a drop-down menu affordance nested inside the text box that will open the auto-complete and perform an initial search. Then, inside the auto-complete, I would display a message like “continue typing to see more matches”.

The names of the entities are very diverse (for example, part numbers where part names may begin with letters or numbers) and so the user may not know what to begin typing for the auto-complete. So the drop-down list helps to get the user started.

Finally, in some cases, the number of items may be small enough to completely display in a single auto-complete list. So in that case, the help text would not be shown.

I do, however, find autocomplete useful in Vista’s file open/save/etc dialog box. The other day I wanted to change the icon for a shortcut I created and knew the name of a .dll file I wanted to open and new it was in C:\windows\something\or\other.

The problem was that the directory in question contained, as you might expect, a multitude of files which would take some time to scan through to find the right one. Instead of scrolling through the list of files and selecting what I wanted, I just started typing in the file name and selected from the autocomplete selection – just two characters and a click did it.

I’d say that autocomplete is a bad feature if you know what you’re looking for as it’s quicker to type in the exact phrase, but can provide benefits if you do know what you want but typing it is a bit tricky, like an obscure file name or an email address.

As an aside, there’s a bit of a mess up with the last but one ‘Related info:’ links: “Bill Scott: Distracting or Narrowing: Looking a Little More At Live SearchD” – only the D on ‘Distracting’ and the odd D after ‘Search’ are linked and incorrectly so.

With the Google Suggest example: it is the only field on the screen. I’d wonder at pushing the interaction too far to increase intuitiveness, where learnability would likely suffice. The only thing I can think of that would better communicate to the user what will/does happen with auto-complete would be to allow the user to turn it on and off (with associated explanatory text next to the enabling widget).

Perhaps the fact that it just happens is part of the problem. Or maybe it’s what we call it. In Google’s case it’s live “did you mean…” Perhaps if framed in a way that most users of Google are familiar with and rely upon, it would be easier for people to use?

Just random thoughts on a Saturday afternoon. :)

BTW: I many cases it is actually faster to simply type out the full word, than to use an autocomplete feature.