Thursday, May 21 2009
So much for search volume as a pop culture predictor. Based on the level of search activity leading up to the American Idol finale this week, a number of online oracles projected Adam Lambert as the landslide winner of the world’s biggest singing contest. “If Americans’ search behavior is any indication which way they’re leaning towards voting for the next American Idol, it looks like it’ll be a runaway victory for Adam Lambert,” wrote comScore analyst Eli Goodman in a Tuesday blog post. It contained a chart showing Lambert boasting a 78% share of all searches among the final three contestants. Hitwise also gave Lambert the nod.
But the victory by perceived underdog Kris Allen Wednesday night turned that logic on its ear. While the final tally wasn’t given, the competition drew a record 100 million votes. Not everyone blew the call, however. OneRiot, which bills itself as a “real-time” search engine, boldly issued an email Wednesday afternoon correctly picking Allen as the winner based on social buzz.
In a blog post, the startup tweaked the search activity-based speculation of other forecasters: “Instead of looking solely at searches, our team was watching what people were actually saying about Idol’s two final contestants on the realtime Web – and they seemed to be telling a different story.”
In short, user data that Boulder, Colo.-based OneRiot analyzes from Twitter, Digg and other social sites showed that Allen was generating more positive comments than Lambert. “Overall, while people seem to talk about both candidates equally as much, it looks like they’re much more likely to talk favorably about Kris Allen, using words like ‘love’ and ‘must win.’ Predictably, this indicates that they will really vote for him in the final battle,” according to OneRiot.
What about all the searches for Lambert? “They might not mean that much, because the actual sentiment of the social web is predicting a different winner,” the company argued in its blog.
On Thursday, OneRiot CEO Alessio Signorini was enjoying a bit of a victory lap of his own. “It was fun last night. There are many fans of American Idol here in the team and they made fun of me all week for this prediction, but then everybody called me last night to apologize,” he said.
Signorini went on to say that search volume can be a good predictor for certain trends. He mentioned Google’s Flu Trends tracking-site as an example. “People do more flu-related searches when they are feeling sick, and if you add to that IP-based geo-localization, you have a pretty good monitor of the ‘health of the U.S.,’” he said.
As part of the Computational Epidemiology Group at the University of Iowa, Signorini himself is involved in applying some of the company’s techniques to weightier matters than guessing the next American Idol winner.
“I personally pursue some projects (Social Web Information Monitoring or SWIM) which try to detect and predict diseases outbreaks, along with their respective public response and sentiment, through the analysis of Web social activities,” he said.
To be fair to comScore’s Goodman, he did point out Tuesday that support was building for Allen, who had narrowed the search gap with third finalist Danny Gokey. “Certainly an interesting indicator in hindsight considering that the dark horse leapfrogged Gokey to make the final two,” he wrote.
Apparently, Web data crunchers weren’t the only ones surprised by the contest’s outcome. After being told by host Ryan Seacrest that he had won, Allen said, “Adam deserves this.” Some cultural critics noted that Allen was actually consistent with the more conventional winners of the past than the brash, eyeliner-wearing Lambert.
“America chose sweetness over sizzle, small-town reticence over Vegas swagger,” concluded The New York Times’ Alessandra Stanley.