Micro-targeting gives a virtually complete picture of every individual, the ability to predict their support for a particular party and, crucially, the issues, messages and mediums that move them to vote. Melanie Batley reports

Motorway Man may well swing this year's general election, if recent reports are right. According to a poll commissioned by Experian, the datamining agency, this group comprises 15 per cent of the population, and the party that woos them is likely to form the next government.

The challenge in every election is to convince undecided voters. As early as 1867, Walter Bagehot identified a group that is perhaps the first recorded ancestor of Motorway Man. He wrote: "Public opinion nowadays is the opinion of the bald-headed man at the back of the omnibus."

What we haven't been told before is how, exactly, to pinpoint the specific messages that motivate these people to come out for a certain party, or to come out and vote at all. But that's where things are changing.

For campaigners, a groundbreaking tool has emerged from the enormous advances in information technology, and it is able to offer some of the answers that elude the pollsters. The tool is called micro-targeting. It enables campaigns to obtain a virtually complete picture of every individual in the population, along with an ability to predict with accuracy their likelihood of supporting a particular party, and, the issues, messages and mediums that will move them to vote.

So while the marketers can tell us that Motorway Man is out there, microtargeting allows campaigns to find these specific individuals and influence them to vote in a particular way.

It starts with the accumulation of massive amounts of personal information to build a so-called political DNA of every individual in the population - only now possible with recent advances in software and data accumulation.

Information is gathered from every publicly available source, including the electoral roll and census, to privately held data such as consumer behaviour, financial history, polling research, and even magazine subscriptions and web-surfing habits. One Washington firm said it uses over 600 different data sources, amalgamating billions of individual records.

Computer models are then applied to the data to create profiles or, in marketing speak, ‘segmentations', of voters that share common characteristics. Sounds a bit spooky, but direct marketing agencies and retail outlets have known these details about us for years. The innovation for the political world is that the data is combined with traditional voting files and digitally analysed to predict with remarkable accuracy who voters are minded to support, how likely they are to vote, and what issues might influence them to vote for a certain party or candidate.

Not only that, but for individuals who may have no voter history, or about whom information is incomplete, the computer modelling extrapolates their political affiliation and voting propensity by comparing their profile to other similar individuals. This is an especially powerful application of the technology, potentially opening rich new sources of voters.

From there, different models are overlaid. A partisanship model would score everyone in the country from one to 100 on strength of party sympathy. Meanwhile, a voter-propensity model would assign a likelihood-to-vote coefficient. Finally, an issue model could be applied to the other two models to pinpoint which messages are likely to influence targeted voters, and how best to approach them. This approach is simply light years ahead of traditional targeting, which relies on voter history, party affiliation and geography.

Using the traditional approach, a campaign would take the list of voters from the electoral roll, sort it by voter history, and compare that with records on party affiliation. With that, it would take a look at the geographical concentrations of supporters most likely to vote, then carpet bomb the areas with literature, door-knocking, and get-out-the-vote efforts. It was leaving vast numbers of potential voters untapped.

What about the people with no voting history but who could be persuaded to vote if contacted about an issue they cared about? How about supporters living in areas considered an opponent's territory? There was previously no way to pick them out. The traditional method also wasted resources on voters who might not vote or, worse, potentially energised people who might even vote the other way.

How is micro-targeting used in practice? The data is applied to voter contact plans to enable campaigns to send individually targeted messages to potential voters, delivered through the mediums most likely to influence them.

By doing this, micro-targeting grabs those undecided swing voters, as well as those less likely to vote otherwise, by talking to them about the things they care about in the way they will listen, and persuades them to vote.

Now this is the stuff that can really swing elections. The most spectacular example of its potency was during the 2008 Obama presidential campaign. Of the 49 million adults targeted through the micro-targeting programme, about half voted, representing more than 20 per cent of all the votes cast in the election. Perhaps more significantly, 82 per cent of the work occurred in 16 swing states, accounting for 37 per cent of all votes cast in these states.

While micro-targeting may be all the rage, it can't win elections. It's a tool, a tactical weapon to support a campaign's voter-contact strategy. Only human beings can interpret how to use that information in the context of the other macro factors in the race and the campaign's resources. Nonetheless, the technology promises to permanently transform the way we campaign, offering greater precision, and potentially stronger engagement with voters; a welcome development in an age of voter apathy.

British political parties are tightlipped about the tools they are usingduring this campaign, but each of the three main parties is already using social segmentation data. The party that best integrates these techniques into its strategy will have an advantage that could swing the election.

Data wizards are already working on the next wave of innovations, from real-time updates on voter views, to ‘relationship modelling' to see who might have the greatest influence on voting behaviour within a household, and even facial recognition software that can analyse facial reactions of individuals in a crowd and link it to the data in their voter profile.

Micro-targeting, it seems, is changing the face of political campaigning in the 21st century. Those bald-headed men at the back of the bus and sales reps ploughing the fast lane of the M6 are no longer an anonymous mass. They are individuals, and they are being watched.

Melanie Batley has worked as a US political operative and as a Conservative Party researcher