We recently had the opportunity to apply market research and marketing analytics techniques doing some pro-bono work for a local political campaign. Apart from the satisfaction of being able to contribute our professional skills to a local cause we were enthusiastic about, this was a great example of how the availability of low-cost, online services are making analytics more and more available to a wider range of applications and budgets.
Situation
We faced an election for a village president and three village trustees. Often these races in this small town of 18,000 people are not even contested, but this year three candidates were running for the president position and five were running for the three trustee positions. We were working for four candidates who were running as a slate for the four open positions.
To our knowledge, formal voter polls and surveys had never been done in this community. Campaigns were shaped by candidates’ informal sense of the community and local issues and outcomes, while often easily guessed at, were not really known until election night.
However, the number of candidates in this election created some interesting campaign strategy questions and issues that led us to conduct a poll and analyze the data because the large number of candidate combinations voters could consider. Specifically:
• Could our presidential candidate afford to ignore one of his opponents to focus his energies and voter attention?
• Should we concentrate on one or two of our trustee candidates because we would be guaranteed winning at least one trustee seat?
• How well would running as a slate work? Can some candidates “carry” others?
• What was the profile of voters with whom each candidate was strong or weak?
• Which were the highest-potential pockets of voters?
Polling Voters: Resources and Approach
We decided to conduct an automated, IVR phone survey. Despite the potential weariness from extensive robocalls during last years’ national election, this was the only cost-effective solution for this campaign. (A column by Carl Bialik at the Wall Street Journal last year provided a nice overview of industry views of using these polling methods in political contests.)
Obtaining our list was fairly straightforward. We were able to append phone numbers to names and addresses from registered voter lists obtained from county election commissions. Note that because the calling was for political purposes, we were not subject to DNC restrictions, giving us a much higher yield of appended numbers than otherwise.
To conduct the poll, we used IVR survey capabilities provided by Ifbyphone. While Ifbyphone provides good text-to-voice capabilities, we recorded the survey to get higher completion rates. We found setting up, scheduling, and accessing survey data from the Ifbyphone system to be fairly easy and straightforward.
We also used Ifbyphone to send out a broadcast get-out-the-vote message the day before the election. We were able to target the message at the household level based on results from the poll, and again the system worked flawlessly for us.
Interesting Learnings
From a marketing and analytics perspective, interesting findings were:
• During a test run, we found that completion rates were terrible in the 3 p.m. to 5 p.m. window. We attributed this to post-school distractions in this bedroom community outside of Chicago. We therefore scheduled the calls from 9 a.m. to 3 p.m. and 5 p.m. to 8 p.m.
• Roughly 45% of the calls were answered, and the completion rate on answered calls was 17%.
• External costs for the entire effort – appended phone numbers, the phone survey, and the broadcast voice message – totaled under $1000. There are small firms that specialize in executing political polls; the cost using one of these firms would have been $1500+ just for the poll.
Results
The poll results told us our president candidate had a 14 % point lead over the principal opponent, with the third candidate a very distant third. Our trustee candidates were in the first, third, and fifth spots, with a negligible gap between numbers three and four.
These findings helped us increase focus on the two weaker candidates in the trustee race and minimizing risks that might jeopardize the lead our presidential candidate held. We were also able to identify the profile of voters where we could make the most headway and infer messages that would resonate best with those voters.
With this intelligence and focus, gained with two weeks left in the campaign, we were able to pull off a clean sweep on election night. More details on the race and specific results are reported by the local press, Suburban Life and the Hinsdale Doings.
Take Aways
For business marketers, we like the possibilities of using IVR polling to obtain quick intelligence on focused questions for a low-cost and with a very quick turnaround (these can be executed within hours). This might be most productively done with customers, who presumably would be more likely to participate and complete the survey. Services like Ifbyphone also create an option for bringing routine customer satisfaction or net promoter surveys in house, saving money or enabling more regular or broader sampling.
Comments