Another big election, another big flop for the pre-election polls. That's what happened in Great Britain, where all the polls badly overestimated Labour Party support and embarrassingly underestimated Conservative votes. The result is that Prime Minister David Cameron has won an outright majority for another term in office when all the experts predicted the best he could hope for was to cobble together a broader coalition.
Why should you care about this if you're not into British politics? Because this polling failure in the UK is just the latest election where the professional prognosticators got it wrong, really wrong. The big pollsters also really blew the projections for the Israeli election earlier this year and didn't do all that well in a lot of the 2014 midterm elections in the US either. Remember, they're called "professional" for a reason; they get paid for this information no matter how inaccurate it turns out to be. And that means we're looking at a very good business opportunity for the firms who can reverse this trend and get this kind of polling right.
How good of a business opportunity is it? The official industrial trade publications say the polling, market research, and opinion tracking business brings in $40 billion annually worldwide. $15 billion of that is in the U.S. alone. And with more key elections on the horizon, and generational shifts throwing traditional consumer tastes for a loop, politicians and the private sector alike are desperate for more accurate polling and marketplace data.
For a while, it looked like the big winner filling that need was Nate Silver and his 538 polling aggregation website. ESPN bought the site and put considerable resources behind Silver, known as a kind of wunderkind for his dead-on accurate prediction of the 2012 electoral college breakdown. As you might expect from the ESPN ownership, 538 does a lot of sports-related data in addition to election prediction and analysis. But that's not unusual in any way because while we might know outfits like 538, Gallup, Rasmussen etc. for their political polling, they also make big money providing predictions and research for purely private sector clients. For them, elections are often just a big commercial where they hope to show off their research chops to potential private sector customers. The problem is, wunderkind Silver and his cohorts at 538 got this UK election wrong too. Meanwhile, the British bookmaker Ladbrokes says an unidentified man bet 30,000 pounds on the Conservative Party winning a parliamentary majority at 7-1 odds and he's going to collect that 210,000 pound haul. Maybe some politicians and ESPN should hire that guy instead.
But I don't say this to trash 538, or the other polling/research companies and what they do. I say it to point out that this is a very tough business with shifting challenges and it's simply foolish to anoint anyone as a permanent savior or all-knowing guru. And while the media might like to make stars out of prognosticators, businesses and serious politicians have to always be ready to hedge their bets. That's one reason why even CNBC recently announced we're going to stop using Nielsen to measure our daytime audience.
While some political pundits and CEO's might think the bad data is the result of some kind of political or cultural bias, the cause of all of this is more likely the fact that the way people in all countries communicate, shop, and live is changing... and fast. For example, any pollster will tell you that data based on a person with a mobile phone is less reliable than someone contacted and surveyed on a land line. And with more and more people shifting over to mobile phones exclusively, that's just one new challenge.
And don't blame this on technology, because it's not the tech-generated data that's to blame. It's more the fault of the humans picking and choosing which data to read and emphasize. Remember that people like Nate Silver don't do the polling themselves, they look at a lot of different polls and try to choose the best ones. And also remember that tracking software and other innovations like it are better than ever at predicting the personal preferences and even political decisions of the people they're tracking. But the big key to polling and accurate market research is being able to chose which individual people best represent the biggest number of other people. Anyone can take the time to call or look at the tracking data for thousands and thousands of voters or shoppers, but besides being tedious that would be too expensive and take too much time. Polls and focus groups are all about getting that solid representative sample. And it's getting harder to chose that sample from an ever-growing pile of technology-produced data.
Another reason not to be too harsh on the polling machines or the humans who program them is the fact that younger Americans seem to be blowing up the traditional consumer trend models and even voting traditions right now. I'm not saying all those things are bad. Some of the things people in the Millennial generation are doing are really good, like showing a better acceptance for changing jobs and saving money better than their Gen X'er elders. But their cooler response to things like buying homes and having children is going to force data researchers to rip up a lot of their consumer and voter prediction playbook. I don't envy them their task.
Again, this is all a way of saying that the prognostication business is wide open for some disruptors who find a way to do this better. It may come from a company that makes polling data that's more "human proof." Or it may come from a company that finds an entirely new way to poll that transcends just asking questions. If I knew how to do any of those things, I wouldn't just write about it, I'd sell it! And so would you because with $40 billion up for grabs in a normal year and likely much more at stake next year with the upcoming presidential election, some very tempting bait is in place.