Andrew Kennedy is the Group Agent & Campaign Director in West Kent. He blogs at

“Put number 17 down as ‘Conservative’” said an enthusiastic canvasser during a recent by-election. I looked at him sceptically, and said: “But nobody came to the door!” “Yes, but they had a Jaguar on the drive, so they must be Conservative.” Having reminded the canvasser that John Prescott had two Jags, I left the VI box blank, and moved on to the next house.

How we engage with voters on the doorstep, the questions we ask and how that data is used during an election must change to reflect the society within which we now campaign.

For most of us, canvassing has traditionally been a simple process of asking a voter, “May we count on your support?”, and recording their intention by a simple code on a canvass card – and then reminding “our people” to vote on Polling Day. For a system developed in an era when politics was a binary choice, this worked well. For the eight or so elections after the war, the two main parties shared around 90 per cent of the votes between them: if you weren’t for us, you were agin us.

But times have changed.  The modern campaigner must address two fundamental changes in the political landscape.

There is now more choice – with more parties competing across the UK, often with regional variations, and the emergence of single-issue and independent candidates. Increasingly voters are willing to trade their support, often identifying with one party nationally, whilst happily voting differently at European, mayoral and devolved elections.

Even in my small corner of Kent, there is clear evidence of vote-splitting due to local circumstances. For example, in the last general election, in one local constituency the Conservative local government vote share was higher than the Parliamentary vote share, while in the other four constituencies it was the other way around. And, in one borough, our local share was a full 20 per cent behind the parliamentary percentage. Until we know why one in five Conservatives failed to support their borough council candidate, we will never make the progress we need to make in the borough – but before we can rectify, we need the tools and the skills to identify.

Canvassing over the years has evolved, but too slowly, as has the training we give to our canvassers. Too many people we send to the doorstep still believe they are embarking on an evangelistic mission, to find or make new converts – whilst, in reality, we are simply gathering raw data. And just as voters’ political allegiances have become more complex, so our methods of gathering data must evolve too.

I first encountered the new CCHQ canvass script during the Clacton by-election. And hated it. It seemed over-complicated and under-explained. A simple 30-second conversation on a doorstep had morphed into a five-minute double-page script with show-cards, and we were expected to interview each member of the household separately. In the heat of a by-election – particularly that one – no-one had the time to explain to the canvassers why we were doing this, or how it worked, or why it was important. Not being told the reasons led to a lack of confidence and even a degree of resentment and hostility, which clearly came across on the doorsteps. This negativity became self-fulfilling, with even the most enthusiastic volunteers translating their antagonism to the respondent. This in turn provided the canvasser with all the evidence they needed that the system was unworkable.

For those who have not used the new script, the simple “May we count on your support?” question, was replaced with, “On a scale of 1 to 10, how likely are you to vote for each of these parties at the forthcoming by-election?”. The score for each party was then recorded on a survey form, along with the answers to a number of other questions on the local and national issues likely to affect their vote.

I made it my business to find out the thinking behind this new script, and exactly how the data could be used to our advantage. Having done so, I immediately understood both the need to change – but also the need to explain that change to our volunteers if we were going to make it a success.

So how does it work, and why?

First, we should not be asking voters a binary question when most want to give a multiple-choice answer. According to some polls, only around 35 per cent of voters are now totally committed to a political party (i.e. will always vote for that party at every election). This explains the significant increase in “Don’t knows” or “Won’t says” – something we seldom heard on the doorstep in the old days.

Second, asking people to rate their likelihood of voting for a specific party on a scale of 1 to 10 is far less intrusive than asking them for which party they are going to vote, particularly if the voter is not fully committed, or is unprepared to self-identify.

Having developed a script which produces a more accurate reflection of voters’ intentions, we also need to understand how to use this nuanced data as a campaign tool. For me, the ability to identify the second preferences of other parties’ supporters is increasingly important in closely-fought elections. Below are two examples of how I have used the data in West Kent.

Last year, we fought a local government by-election which was a four-way contest between Conservative, Labour, Liberal Democrat and Green, with UKIP and an independent candidate muddying the waters too. By knowing which UKIP and independent voters would choose Conservative as a second preference, we were able to target our Get Out The Vote material to specific voters – whilst ignoring other UKIP voters who would have preferred a Labour councillor. And by recording the issues which motivate those voters, we were able to produce voter-specific pledge letters, often resulting in two or three different letters being sent to voters within the same household.

The second way this new system helped was to identify Conservative pledges who were seriously considering voting for another party. These voters also received targeted mail, dealing with the issues important to them, just as it provided our candidate the opportunity to spend his time talking to the swing-voters who would decide the outcome of the election. Our vote share was only 27% per cent, but we won with a majority of 53 votes. Without the data provided by this new canvassing method, and the ability to target specific voters, we would not have won.

As too often happens, changes are “handed down” without the training and support needed, nor any explanation as to why these changes have been made. Like most Conservatives, I am resistant to change for change’s sake, but once I am convinced of the reasons why, I become the most enthusiastic convert.

You can, it turns out, teach an old dog new tricks – but he needs a reward. For the dog it is a tasty titbit; for us, it is victory.