PDF
Database Party: The 2002 Leadership Campaign for the Canadian Alliance
Tom Flanagan

This article describes the use of technology in the successful campaign of Stephen Harper to become leader of the Canadian Alliance.  Of course, there is much more to a campaign than technology.  Candidates travel almost non-stop meeting and speaking to thousands of party members.  The mass media has a role in forming opinions about candidates.  Nevertheless the author argues that database technology can be decisive in determining the winner.

For more than two centuries, information technology has steadily reshaped the way politics is conducted.  Members of Parliament were originally elected by public ballot at small meetings held on the village green or in the local tavern.  The progress of print technology in the nineteenth century made universal suffrage workable, once newspapers, pamphlets, and posters could carry political information to a mass audience.  The trend toward mass media intensified in the twentieth century as the rise of radio and television made it possible to address huge audiences with little or no time delay.

A somewhat different trend, however, emerged towards the end of the twentieth century, as the electronic computer once again revolutionized communications technology.  Using inexpensive personal computers and database software, it became possible to address large numbers of people and to keep track of their preferences as individuals. The result was to restore some of the characteristics of face-to-face politics through the impersonal but individualized media of direct mail, phone banks, Internet surveys, and e-mail and fax blast outs.

Each of these innovations in communications technology has supplemented rather than replaced its forebears.  As a consequence, anyone doing politics today has to integrate a wide variety of communications channels ranging from face-to-face contact, through print and electronic publication, to the most update manifestations of database technology.  In such a complex situation, there is no single right way to do things.  Rather, the combination of media chosen will depend on the nature of the contest and the resources available to the contestants.

Rules and Resources

The Canadian Alliance leadership campaign was a one-person, one-vote contest, in which every registered party member had the right to vote through a mail ballot.  The Alliance leadership contest rules made no attempt to weight constituencies equally; rather, they weighted members equally.  Unlike Liberal and Progressive Conservative leadership races, four thousand votes in the constituency of Red Deer were worth no more and no less than the sum of 400 votes in 10 different ridings.  Such rules put a premium on communications media that could reach people as individuals, rather than deal with them as bunches in ridings.  Paid advertising was too expensive to use because the party’s members were spread fairly evenly across the vast area of Ontario and the four western provinces, with a much smaller number in Quebec and Atlantic Canada.  We had to put information in members’ hands or speak to them individually if we hoped to capture their allegiance.  Our resource constraint was that we had very little money when the campaign began, so we had to find a communications strategy that did not involve large upfront costs and would generate revenue as we went along.

Raising Money

In the so-called “Draft Harper” phase of the campaign (August-November 2001), Stephen Harper travelled, met people, and used the telephone to ask for personal and financial support.  We sent out some letters and put up a website, which brought in a few donations and offers to work on the campaign; but this was a minimal use of database technology.

Everything changed in early December, when he officially registered as a candidate.  Paying the $25,000 deposit and meeting certain other requirements gave us access to the Canadian Alliance membership database, which, under the circumstances, was the equivalent of an Elections Canada voters list in a general election.  The party’s database contained about 300,000 names, including anyone who had ever been a member since the foundation of the Reform Party in 1987.  Of these, about 70,000 were current members as of December 2001.  The database contained phone numbers and mailing addresses, which were generally reliable for recent members but became less reliable the longer the date since the last renewal.  In a small but still useful percentage of instances, the database also contained fax numbers and/or e-mail addresses.

Our first and urgent need was to raise money to finance the rest of the campaign.  We began, therefore, by sending a direct mail piece to the 70,000 current members, asking for their financial support.  This and subsequent mailings set up a stream of income that funded the campaign adequately.  In the end, we raised about $1.1 million from 9500 donors, for an average donation of $116.  Some donations came at rallies, or at special dinners, luncheons, and receptions, or over the Internet, but by far the largest share resulted from our direct mail program.

Once it was obvious that direct mail was working for us, we sent out a second wave of letters to lapsed members, urging them to rejoin the party to vote for Stephen Harper, and, of course, to give money to support the campaign.  Here we had to make a strategic choice because the list of lapsed members was too large to mail everyone on it.  We chose not to approach at this time the very large number of names (over 100,000) who had joined in 2000 at the formation of the Canadian Alliance but had not renewed their membership in 2001.  Rightly or wrongly, we assumed that these potential voters would be less receptive to our message of revitalizing the Canadian Alliance but might be more receptive to other candidates who were at the time emphasizing cooperation or merger with the Progressive Conservatives.  Instead, we went after people who had joined the Reform Party before 2000 but had now let their membership lapse.  (We did not bother with people who had not renewed since 1995 or earlier, believing that too many of them would have died, moved, or lost interest to make a mailing worthwhile.)  That still left us with a list of about 90,000 names for the second wave.

This second wave was also a great success.  It not only paid for itself several times over in contributions, it brought in over 2000 membership renewals, i.e., people who would now be able to vote for Harper.

Voter ID

We set up phone banks in January 2000 to move forward with Voter ID.  Our goal was to contact all eligible voters to see if they were supporting Harper or one of the other three candidates.  We could identify several thousand supporters by databasing our direct-mail and Internet donors, and we also identified another thousand or so by means of cost-free e-mail and fax blast outs.  But the main job of Voter ID had to be done over the telephone.

Though money was starting to pour in, we did not feel we could afford to hire a commercial telemarketing company.  Nor did we want to set up a full-scale predictive dialing system with callers using computer terminals to record answers to issue questions and to renew memberships and take donations via credit cards.  Such systems require well trained and paid staff, whereas we wanted to use the dozens of volunteers that were available to us in Calgary and, to a lesser extent, in Ottawa.  We therefore opted for a stripped-down predictive-dialer system in which the call stations consisted simply of normal telephone keypads.  These could not have coped with complex polling questions and financial transactions but were adequate for coding the answer to a simple question about which candidate the respondent supported.  We set up one bank with 20 seats in Calgary and later another similar bank in Ottawa.  Because of the simplicity, we were able to make extensive use of volunteers, though we also manned some shifts with teen-agers who were paid for their time.

Throughout January and February, our phone banks worked systematically through the party’s database, calling everyone who would answer the phone.  Those coded as undecided got a second call late in the campaign to try to persuade them to come over to the Harper side.

Membership Sales

Although our fundraising and Voter ID efforts were going well, it became apparent in the second half of January that our campaign plan had one serious shortcoming: we had not put enough emphasis on getting renewals and selling new memberships.  We were winning the battle of persuasion within the voters’ list of current members, but there was a chance that Stockwell Day could defeat us by using sales to enlarge the pool of eligible voters.  Having just been through a hard-fought leadership race in 2000, the Day camp was familiar with various sales techniques and could hope to get renewals among people to whom they had sold memberships in that year.  We were selling memberships at Stephen’s rallies, and our second wave of direct mail had also brought in a couple thousand renewals; but these efforts were not nearly enough in light of the sales numbers that were starting to trickle out of the Day organization.  We simply had to ramp up sales and renewals, and we had only one month to do it, because sales had to be entered in the party’s main database before March 1 in order to confer eligibility to vote.

We had the support of almost 30 members of caucus by this time, so we started to press our MP supporters to mobilize their local organizations to sell memberships in their ridings.  We also placed some newspaper and magazine advertisements urging readers to phone, fax, or e-mail memberships and donations to us.  Beyond that, we tried to think of ways to use our database for sales purposes.  We were now reasonably well funded, so we could afford to hire a telemarketing company to make about 25,000 calls to lapsed members in ridings where we could not count on the support of the local MP.  We also thought of two ways to exploit auto-dialer (“demon dialer”) technology.

Our first experiment was to have Stephen record a message saying more or less,  “Hi, this is Stephen Harper.  If you want to rejoin the party and support me, press 1.”  Using an American auto-dialer with a thousand ports, we were able in an hour to send this message to the more than 100,000 lapsed members that we had not yet mailed because they had joined during the formation of the Canadian Alliance.  About 1700 people pressed 1, and the computer spit out a list of their phone numbers for call-backs.  It turned out that not everyone who pressed 1 really wanted to rejoin; some were just trying to make the message go away.  But we still got several hundred renewals out of the exercise.

An even more successful and cost-effective experiment with the auto-dialer involved the approximately 30,000 Harper supporters that we had identified and databased by mid-February.  We sent them the following message recorded by Stephen: “It’s a very tight race and every vote will count.  I need your help.  Please sell some memberships as quickly as possible and phone them into the campaign office.”  Almost as soon as Stephen’s supporters got that message, the phones started ringing incessantly in the office as supporters called in the sales they were making to their family members and friends.

In the end, we sold about 16,000 new memberships and renewals.  Stockwell Day’s campaign beat us handily in that department, claiming sales and renewals of 29,000; but their advantage in sales was not enough to overcome the big lead we had built up among current members.  However, if we had not gone all out to sell memberships in February (and database technology was a big part of what we did), we could have lost to Mr. Day.  I had a lot sleepless nights that month until the sales campaign was over!

GOTV

The final step in our communications plan was a “Get Out The Vote” (GOTV) message, for which we again resorted to database technology.   In late February we sent a letter with a new pamphlet to all current members except those we had identified as supporting other candidates (we didn’t want to encourage them to vote!).  We also sent a letter to all the new members sold by the other campaigns after we received those updates to our database in early March.  We did not try to pry them away from their first choice, but we asked for their second-ballot support if no one got a majority on the first ballot.  Both of these letters also contained a mild request for financial support, which helped keep up the flow of contributions required to cover late campaign bills.  Finally, we used our phone banks to call all our identified supporters and remind them to vote.  We could have done this with the auto-dialer, but our volunteers and paid teenagers were still keen, and we thought a live voice would be more effective than a recorded message for GOTV.

Results

We succeeded in saturating our core supporters with messages.  Most of our main target group—current members—received at least two phone calls and two letters from us during the campaign, plus auto-dialed announcements of campaign events in their area (donors also received additional thank-you letters with special enclosures).  Others on the list, depending on which category they fell into, received varying combinations of auto-dialed messages, live calls, and direct mail.  A few members rebelled at all this communication and said emphatically, “Stop bothering me!”  Far more people, however, seemed to relish the contact, regarding it as a sign of life in a party they still cared deeply about.  Many prolonged their phone conversations, seeking to find out more about Stephen and his views.

Although I cannot offer conclusive proof, my experience on the campaign leads me to two conclusions: First, our voter-contact program enabled the Harper campaign to win on the first ballot.  A small poll we commissioned before the campaign showed Stephen Harper and Stockwell Day in a virtual dead heat, each receiving the support of 20-25% of current members.  But Harper wound up getting an absolute majority of 55% on the first ballot, compared to Day’s 38%, even though Day sold many more memberships.  This means we must have won the support of an overwhelming majority of current members—and they were the people we saturated with letters and phone calls.

Second, our voter-contact program helped bring about unusually high turnout in this contest.  Almost 71% of eligible voters cast first-round ballots, compared to 60% in the first round and 56% in the second round of the 2000 leadership race, and 42%, 50%, and 66% in the party referendums of 1991, 1999, and 2000.  Our letters and phone calls helped energize members and get them re-involved in party affairs.

Together, these two results contributed greatly to helping Harper get off to a successful start as Leader.  Winning on the first ballot meant that the leadership issue was closed.  Stockwell Day, though not humiliated, had lost decisively, and his followers had little choice except to rally round the new Leader.  In contrast, a second ballot would have prolonged and perhaps exacerbated the internal factionalism that had been tearing the party apart for over a year.  Also, members’ renewed interest in party affairs led to increased financial contributions, which put the party back in the black and enabled it to pay off a $2 million bank debt in the fall of 2002.

It is fitting that database technology helped revive the party, because the Reform Party from the beginning had always been based on database technology.  The founders of the party had the great good foresight to insist on keeping a single membership list in one database in the national office.  They thus avoided the problems plaguing other federal parties whose lists are kept by provincial wings.  The single database has sustained the Reform Party and Canadian Alliance, in good times and bad, by allowing the national office to conduct a unified program of internal communications and fundraising within the membership.  Founded in 1987 when the new era of communications technology was just starting to come into its own, the Reform Party (and the Canadian Alliance as its successor) is truly a database party.