I’m posting pieces of the book once a week, but in between I’ll occasionally post a comment on a related subject. Today’s topic is market research and the television show American Idol.

If you live in the US you’ve probably heard of American Idol, especially if like me you have a pre-teen daughter who asks you to call in votes for her favorite singers after bed-time. If you don’t live in the US, chances are you have a local version of the same show – Croatian Idol or Ugandan Idol or something like that.¹

In case you’ve been living in a cave or don’t have kids, here’s how the show works: A group of amateur singers performs every week, one song per performer. The audience votes, and the night after the performance, the person who received the fewest votes has to leave the show. It sounds deceptively simple, but the personalities involved are entertaining, and if you watch the show regularly you kind of get attached to the singers and it’s a wrench when one of them gets voted off and you realize you won’t ever get to see them sing again.

Mind you, I know all of this only from watching my daughter. I myself don’t pay much attention to the show, oooooh no, I’m too busy writing blog entries.

The central tension each week is the mystery of who will get voted off. Or anyway, that was the central mystery until a few weeks ago, when an online tool called Dial Idol hit its prime.

Dial Idol is a website and a software program. You install it on your computer and use it to dial your votes into the show, via modem. The program also reports your votes back to the website. This isn’t all that interesting – there are lots of American Idol polls online. But Dial Idol also tabulates the percent of the calls for each singer that generate busy signals. American Idol gets so flooded with calls during the voting process that it’s commonplace to get a busy signal – you might have to call three times to cast a single vote.

The genius of Dial Idol is the use of the busy signal. Any online poll that people can volunteer to take is plagued by self-selection errors. But the ratio of busy signals to total calls turns out to be an accurate predictor of who’s getting the most votes. It corrects for any bias in the people choosing to take the poll.

The website has been in operation for years, but just recently it reached some sort of critical mass. I don’t know if it was the total number of people using the software, or tweaks they did to their formulae, but the site is now turning out eerily accurate predictions of the outcomes of the voting. It has correctly predicted the people voted off for the last four weeks straight.

You may well be thinking, who cares? And in one sense this is all trivia. But to me, Dial Idol is a great example of the sort of interesting market research that’s being enabled by the Internet. We’re in a golden age of new market research techniques. They’re giving us more ways to understand people, at lower cost, than we’ve ever had before. Some day we’re going to look back at the days before the Web and wonder how we ever managed to do any marketing at all.

1. I thought I was joking when I wrote that, but then I looked it up and it turns out there is a Croatian Idol, called Hrvatski Idol. No sign of an Idol program in Uganda, although there is one in South Africa. You can see the full list of 32 countries here. (That’s yet another win for Wikipedia over Encyclopedia Britannica, by the way.) [↩ back]

This entry was posted on Thursday, April 20th, 2006 at 12:18 am and is filed under Commentary. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.