Measuring the impact of advertising on the buyer’s journey through the sales funnel has become known, in marketing parlance, as attribution. It is a hot topic among CMOs, agencies and in a growing number of SMB organizations. It is also a topic that will be of keen interest to those agencies that specialize in supporting political campaigns in the lead-up to the 2016 elections.
Last week I was at an ad-tech conference and witnessed a panel of political campaign experts surrender to the fact that an accurate attribution of marketing spend in politics was difficult, if not impossible, to attain. With 85% of the electorate spending time online these days, candidates and campaigns recognize attribution is an issue that needs to be addressed. They know they need to go big across multiple online channels if they are going to be competitive. Every race since 2008 has also taught them that political agencies need to provide more proof that the voters they are targeting are actually being reached, and that the ads that they are running are actually working.
Attribution solutions for non-political marketing efforts have a huge advantage: they know exactly what happens at the end — a buyer makes a purchase in their own name. The value of the advertising that led up to that purchase is determined (often subjectively) through a set of metrics that assign credit to the advertising seen by a buyer before the sale. Assumptions are based on a statistical sample of buyers that is broad enough to develop an attribution pattern through predictive modeling. This allows advertisers to adjust their media spend in order to generate a better return on their marketing investments.
Unfortunately, executing a similar attribution solution in politics is a bigger challenge:
There is no purchase event
Identifying how an individual votes is not possible. Voters may visit a candidate’s site or social page but vote anonymously, which means conversions are nearly impossible to measure. One can get a sense of voter engagement though Twitter or Facebook likes or follows but no one really knows “who” is checking out a candidate online.
Attribution depends on tracking cookies
Only a small percentage (20 - 30%) of voters have an accurate and active cookie associated with them at any given time. In addition, only a very small number (less than a tenth of 1%) will opt-in and follow a candidate online. In any local or state race where less than one-third of the voters are even reachable online, the optimization of online media spend based on a tiny sample of one-third of 1% is statistically risky.
Voter Outreach is Omni-Channel
Campaigns structure voter outreach around fund raising, messaging and get-out-the-vote activities and they use many channels to do these things: call centers, door knocking, email, search, display ads, social media, pre-roll video, television and radio. Using a very small number of conversion samples across multiple media channels to measure attribution is, again, a statistically risky move.
Not too long ago, attribution in politics was straight forward — a piece of direct mail landed in the mail box of the voter and they registered their intent by either filling out the prepaid post card or not. The phone rang (usually in the middle of the dinner hour) and when the call was finished you had pledged your vote or you had agreed to send the candidate some money. Even when TV ads run incessantly from September into November, Nielsen was there to measure and report that you watched them all.
Politic campaigns will continue to expand their investment on digital outreach in the future as long as the attribution results support it. With on-boarding match rates of only 20 - 30% the venerable “cookie” is particularly unsuitable for political campaigns. For digital outreach to perform most effectively in politics, campaigns need to know with whom they are engaging. Alternative identifiers such as IP targeting or a common voter identifier that protects privacy but is able to map across all voters at a local, district or state level are options that are being looked at. How they will be implemented and how they impact the practice of political attribution will be worth watching as the 2016 election approaches.