The short answer to this question is “shorter is better”, with one caveat.  As I mentioned in an earlier post, I think :15 spots are generally a waste of money.  It’s not enough time to develop a real story, and (with some notable exceptions) doesn’t usually communicate much of anything, except the sponsor’s logo.  It’s just long enough to be annoying, but not long enough to entertain, unless you’ve got a fabulous site gag that works in 10 seconds (that’s the exception).

Even :30’s always seemed to be a little tight to me.  I’ve noticed over the years how often the first assembly of a :30 spot comes in around 40 seconds.  Then we would squeezed it down and speed it up to make a :30, but that work usually compromised the communication and truncated the storytelling.  No one would ever know that unless you had seen the first cut, but it’s generally been true.

So my best guess is that videos should be between 45 seconds and 3 minutes.   That’s plenty of time to tell a good story, develop some drama, set up characters and situation for a comedic pay-off, or do a product demonstration.  But not so long as to bore anyone.  The research that’s been done to date says that video completion rates start to fall off fast after 3 minutes.   If the video is really interesting, and moves at a decent pace, you can stretch that to 5 minutes, but you’re tempting the audience to bail out before it’s over.

Mind you, we’re talking here about short web content, not a traditional TV series, with a stable cast and long dramatic arcs.  Those work on the web too, but only with massive amounts of promotion and exposure through traditional media first.  My kids watch “Glee” online all the time (it’s time-shifting for them), and have no trouble sticking with it until the end (although they don’t pay much attention to the commercials).

I mentioned the great documentary series that Honda has done for web distribution a while ago.  Those docs are between 15 and 20 minutes long.  I think they did a great job executing them, and I intended to go back and watch the rest of them.  But I haven’t.  Would I be more inclined to watch them all if they were shorter?  For sure.  Most people don’t have time in their day to spend 20 minutes watching a web video.  But they might do it after work if they considered it pure entertainment.  Again, that’s a different content category.

What to do if you have more than 3 minutes of stuff to say?  I have two suggestions.  First, I have been impressed with the way this is handled at Grovo (http://www.grovo.com).  This is a website that publishes videos meant to educate their audience about web technologies (how to use LinkedIn, for instance).  Many of their courses run 15 or 20 minutes total, but they’ve chosen to break them into smaller chapters that can be viewed individually, each chapter taking 2 or 3 minutes.  It’s a great strategy, because you can watch one or two, then if you get interrupted or bored, stop watching.  Then, when you have time, you can go back and pick up the next chapter in line.  It allows you to skip over sections that you already know or don’t have any interest in, too.  I think it works quite well.

The other strategy is to post multiple version of the content.  I produced a video to promote a local charity last year (babyspace.org), and the video turned out to be around 6 minutes.  We had much discussion about the length, knowing it was problematic.  And in then end, we cut a shorter version to post on the main web site that was around 3 minutes.  Then we also posted the full-length version to YouTube, and included a link to that version on the charity’s main web site.  That way we didn’t burden the casual viewer with content that was too long, but also allowed those who wanted to know more to see the longer version.  Problem solved.

So there you have it.  Avoid content that is too short to be of use, and that which is so long as to be tedious, and all will be well.  It’s always a judgment call, I know, but smart publishers will err on the shorter side.