You poll if you want to... it’s analysis that counts

Outlook

The Independent on Sunday carried out a similar polling exercise in April 2010, in which eight out of eight pollsters predicted a Conservative overall majority (Getty)

When companies put out financial results or trading statements there is almost always a “but” to be found somewhere, however good the numbers look at first glance. Turnover and profit might be up, but what about margins? The business might be growing like a train, but what about that mountain of debt?

YouGov might be one of those rare exceptions that proves the rule. So no buts at all? Not quite. It’s just that the financial effect of it is negligible. I’m talking, of course, about the way YouGov got the general election result completely wrong.

UK election polling provides a tiny fraction of the group’s revenues. But it’s very visible. As YouGov’s trading statement amply demonstrates, companies are still beating a path to its doors. One reason for that might be because the company appears to take its failure rather seriously. It is preparing to release a full-scale post mortem into what went wrong.

The top brass remained coy about what the review might say when I talked to them. But when YouGov asked people which they would rather see in power – a coalition led by David Cameron or one led by Ed Miliband – the former was favoured by a margin of 14 per cent. Which was completely out of step with what people were telling YouGov about their actual voting intentions. The importance of that was missed.

In a world ever more obsessed with surveys, statistics and “big data”, this leads one to ask if things are being missed in the analysis of data provided to companies?

The deep “granular” analysis that YouGov does is dazzlingly sophisticated and can be quite fascinating. Its look at the Germans and their likes and dislikes recently produced a bestselling book, some 27,000 copies of which were sold in just five days. Big data like this can be very seductive, and big business has certainly fallen in love with it (which means big profits for YouGov).

The problem is not so much with the data itself, it is how it is analysed and interpreted. When that falls down, things can go badly wrong, as the election debacle demonstrated.

Occasionally, therefore, it might just serve progressive companies to step outside of it completely.

Customers, after all, are not just statistical constructs that can be conveniently wrapped up and boxed. They are individuals. YouGov’s research would probably tell its clients that they would respond rather well to businesses that recognise that fact.