Home Resources Newsletters Voodoo Analytics: Tactic Attribution

Voodoo Analytics: Tactic Attribution

August 01, 2014

Instead of attempting to attribute pipeline or revenue to individual marketing tactics, apply a holistic touch analysis to successful deals

Under a baking sun, a weary traveler in the desert trudges across a seemingly endless expanse of sand. Suddenly, he spots a sparkling lake in the distance. He rubs his eyes. It’s still there. He gleefully picks up his pace, only to watch the mirage of water melt into thin air.

Weary marketers are searching for effective methods to quantify the impact of their programs and justify the costs of their tactics. In response, marketing technology vendors are promoting features they claim will reveal the revenue or pipeline impact of individual marketing tactics. However, associating one or several tactics with revenue and pipeline is an uncertain process, and the benefits may be a mirage. In this issue of SiriusPerspectives, we explain tactic attribution, share the challenges inherent in its use and misuse, and identify a narrow set of circumstances in which it can be used effectively.

Defining Tactic Attribution

Tactic attribution examines the historical data for multiple deals, either up to the point of their transition into opportunities in the sales pipeline or into closed deals; marketing tactics that touched each deal are identified and assigned a score to indicate their contribution. Individual deals are aggregated into total pipeline or revenue, and the contributions of tactics are similarly aggregated. The result is the perceived contribution of each tactic to pipeline or revenue, which enables a computation of the ROI for the marketing investment in that tactic. The following attribution techniques are commonly used to score marketing tactics:

  • First touch. The initial contact of a prospect with a marketing tactic is assigned the credit for the opportunity or sale. This technique is helpful in identifying the tactic that is most likely to get a new name into the database, which is useful if database expansion is important. However, it is biased toward tactics at the top of the demand waterfall, devaluing late-stage tactics.
  • Last touch. This technique assigns credit for the opportunity or sale to the tactic immediately preceding the transition (e.g. conversion to sales qualified lead or closed/won opportunity). This technique might be helpful in identifying tactics that are best at bringing the buyer over the threshold, but it devalues tactics that initiate marketing engagement.
  • Touch spectrum. This technique assigns weighted credit to different touches in the sequence of interactions. The simplest variety of touch spectrum is linear weighting (e.g. 20 percent of the credit is assigned to each of five touches). More complex approaches may assign variable weights to tactics based on their timing or order. Some models apply extra weight to some tactics (e.g. the first touch, the touch prior to conversion to an opportunity) and an even distribution of weight to others. Some models apply different weights based on contact or buying center roles, and dissipate weights based on time lapses between touches or the age of the contact. Some of these methods are highly sophisticated (e.g. regression, factor analysis).

Problems With Tactic Attribution

Regardless of how sophisticated the analysis of tactics is within the timeline of the deal, the following problems can result in the potential for skewed results and faulty decisionmaking:

  • A lot of a little is still a little. Buying centers are composed of individuals who interact with each other in ways that are too variable and complex to be easily or accurately measured through marketing automation. For example, several members of a technology evaluation team may download a technical document, resulting in an elevated score for that tactic. Meanwhile, another tactic (e.g. senior buyer reads solution brief) that has greater impact on the deal may receive a lower score. Adjusting scoring to account for these kinds of variations can improve accuracy but also introduces subjectivity into the analysis.
  • Inbreeding weakens the gene pool. The top scores are awarded to the best among a subset of possible tactics (e.g. tactics that have already been deployed). The self-referential nature of the analysis reinforces tactical choices that may be suboptimal when compared to a broader pool of unselected alternatives. As a result, tactic attribution analysis encourages the refinement of tactics that have been successful and the elimination of those perceived to have lower impact. Without an infusion of new tactics and an expanding portfolio of alternatives, the list of acceptable tactics becomes shorter, which can lead to stagnation and an overly cautious environment.
  • The past doesn’t always predict the future. Tactic attribution is a form of historical analysis that leads to recommendations based on an assumption that past behavior is an indicator of future behavior. While this may be true in some cases, it is not always valid. For example, an ROI guide may have been effective in the past, but its impact may wane as competitors publish similar content.
  • If I can’t see it, it must not be there. If tactic attribution guides planning and budgeting, marketers risk eliminating or underfunding tactics that are working but are not being measured. For example, user groups are events where valuable advocacy often occurs, but this is difficult to measure. Tactic attribution may score user groups poorly, resulting in their elimination and a negative impact on marketing performance.

Where Tactic Attribution Makes Sense

Tactic attribution can be more reliable in certain circumstances. These include:

  • Simple transactions. If the buying process involves a limited number of participants and interactions and has a short duration, a marketing tactic can be correlated to the sale or conversion. The simpler the process, the more reliable tactic attribution can be.
  • Relationship transactions. If the sales environment involves a master account relationship, and the seller’s marketing efforts include targeted tactics (e.g. special offers or promotions) to drive transactions from multiple buying centers within the customer organization, these tactics can be reliably associated with purchase behavior. Examples include replenishing materials, add-ons, or additional licenses and upgrades.

The Sirius Decision

If you ask marketers who are justifying investment in a trade show, or a bigger slice of the marketing budget, whether they really believe their tactics actually drove the zillions of dollars of business that their tactic attribution analysis claims, they will likely say, “No, but the analysis says it did, and this helps justify my budget and my activities.” Marketing leaders need to avoid creating an environment where shaky analysis leads to a rigged competition for resources and the propagation of activities that “the machine” claims are the best ones. Data-driven decisionmaking in a complex b-to-b environment requires marketers to understand the flaws inherent in voodoo analytics, and shift their focus from justifying individual tactics to exploring which combinations and sequences of marketing touches are predictive of successful outcomes.