Dave the research grouch: Attribution

slicing a pieAttribution seems to be an increasingly popular topic. Recently several networks decided to pilot a new attribution model that includes TV. This effort is intended to head off a number of existing attribution models that only include digital exposures.

The primary problem with either flavor of attribution model is that neither includes all sources of exposure, only those ones that are easiest to measure or that vested interests are willing to pay to measure. Even if we have digital and TV, what about radio (valued for the last exposure when heading to a store), magazines (valued for trust and a platform for presenting detailed ads), sponsorships (affinity with the consumer’s “tribes”), and so forth?

This is a bit reminiscent of so-called “cross-platform media” measures over the past decade, a term that drove me crazy, because most really only covered digital media. And even then, some were only desktop or mobile browsers, no apps included; and forget about inclusion of streaming to digital TV.

A second issue is how these attribution models are presented, which typically is the relative value of each media source’s influence on a purchase. Up to now, this usually meant that the last medium consumed gets the credit, with little or no “attribution” to the stack of other media sources which may have come before. What about the ad that informed a consumer that a product existed, or convinced a consumer of the value of a brand, or gave a consumer the information on a product that would lead to a purchase?

Putting aside the issue of if all relevant media are measured, a larger overarching matter is the seemingly intractable issue of measuring the influence of every exposure on every medium and its relative impact on purchasing – what I would think are needed variables for true attribution. Even for exposure and time spent, some media will have to be respondent-reported, not passively measured. Attentiveness to a property or the ad within it – that’s going to rely on self-report. The trust in, and influence of, each property and its ads, for the specific product of interest – self-reported also. How are these models going to measure all of these aspects for each ad exposure, with so many factors being subjective?

To what I attribute my grouchiness

I’m not naive enough to expect that such a data vacuum will remain empty, even if the solutions offered are questionable in some respects. Solutions are filling this space, and some may be very successful businesses – there have certainly been enough examples of that over the years. My grumble is that the nomenclature used and the way data are presented in these services may not accurately reflect their limits, leading to specious headlines and underinformed users.

David Tice is the principal of TiceVision LLC, a media research consultancy.
Get notifications of new posts – sign up at right or at bottom of this page.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.