At a high level, 2014 has been a year of questioning the productivity of digital marketing and related measurement of success. For example, the most frequent C-level complaint about digital is not having a clear understanding of bottom-line digital impact. For background on this topic, see articles here, here, and here.
I’d guess this general view probably has not been helped by the trade reporting on widespread problems in digital ad delivery and accountability systems, where (depending on who you ask) up to 60% of delivered “impressions” were likely fraudulent in one way or another. People have commented on this problem for years; why it took so long for the industry as a whole to fess up and start taking action on this is an interesting question!
If the trends above continue to play out, over the next 5 years or so we may expect increasing management focus on more accurately defining the contribution of digital – as long as management thinks digital is important to the future of the business.
If the people running companies are having a hard time determining the value of digital to their business, the next logical thought is marketers / analysts probably need to do a better job demonstrating these linkages, yes? Along those lines, I think it would be helpful for both digital marketers and marketing analytics folks to spend some time this year thinking about and working through two of the primary issues driving this situation:
1. Got Causation? How success is measured
In the early days of digital, many people loved quoting the number of “hits” as a success measure. It took a surprisingly long time to convince these same people the number of files downloaded during a page view did not predict business success ;)
Today, we’re pretty good at finding actions that correlate with specific business metrics like visits or sales, but as the old saying goes, correlation does not imply causation.
If we move to a more causal and demonstrable success measurement system, one of the first ideas you will encounter, particularly if there are some serious data scientists around, in the idea of incremental impact or lift. This model is the gold standard for determining cause in much of the scientific community. Personally, I don’t see why with all the data we have access to now, this type of testing is not more widely embraced in digital.