Asking The Tough Questions

Sometimes the holidays provide just enough breathing room to think about important issues that you may not ordinarily have the time to deal with during the normal course of the rat race.

Here are a few of the tough questions that I am recommending marketers think about as we round the corner into a new year of challenges and triumphs.

What’s inside that black box? Programmatic media buying is here to stay. Every ad tech company, and a growing number of media companies, tout a unique algorithmic approach to optimization and consumer profiling -- their version of a better digital media mousetrap. Comparing the technology is virtually impossible. Jump in headfirst and test the partners that feel right. Performance will dictate your ongoing choices. It’s not always the algorithm that wins. Unfortunately you’ll never be able to dig into the differences between one algorithm and another.

What is the best attribution model? The easy answer is that there is none. There is no “best”; there is only “better than now.”  Statistical algorithmic-based modeling is in its early stages, and most attribution systems don’t even offer algorithmic modeling (yet another black box). Make a point to limit arbitrary assumptions and customized weighting. If you question the contribution and weighting of a channel  -- for example, view through retargeting conversion, or branded search -- take the time to isolate and test the contribution of these channels separately, and apply your learnings to the model. Be systematic. Leave the emotions at the door.



What is the impact of [insert tactic here] on my business? Too many marketers and agencies get so caught up in proxy metrics that they fail to see the bigger picture. Have you developed a measurement program to correlate and translate an increase in social media, mobile or other digital marketing engagement into business value? What about branding effectiveness metrics? Nobody’s saying it’s easy.

What story does my data tell? Nobody benefits from data for data’s sake. “Developing insights from data” is such a buzzworthy phrase that it truly hurts every time I say it, but nonetheless a valid mantra. Visualize your data to create a narrative of your efforts over time. Don’t be afraid if the story has some dark chapters; just work towards a happy ending.

How can I improve, personally? Are you doing all you can to be productive rather than busy? Are you a master of communication and collaboration among your colleagues and partners? Do you rely on too many assumptions? Do you admit when you don’t know what you don’t know and look to others for help? We all have some areas where we know we need to improve. Pick at least one and remove it from your list by the end of first quarter.

Every category and marketer possess unique nuances, and our marketing and analytical paths are at varying stages of maturity. I would recommend augmenting the above list with a few burning consumer insight questions related to your specific business. If you had two or three key consumer insights that could change the way you market to consumers online, what would they be? Challenge yourself or your agency to develop a methodology to get there.

What tough questions do you plan to answer this year? Post them in the comments or hit me on Twitter @jasonheller.


2 comments about "Asking The Tough Questions".
Check to receive email when comments are posted.
  1. Kirby Winfield from Dwellable, December 20, 2011 at 1:28 p.m.

    Jason, great article. I think it's telling that your first two points involve a lack of transparency in programmatic buying and attribution. Advertisers should strive to understand and control the inputs and weighting of data that influence pre-bid decisioning - otherwise they're leaving margin on the table for the middlemen.

  2. Jason Heller from AGILITi, December 20, 2011 at 5:17 p.m.

    Thanks for chiming in Kirby. A valid point from a DSP/DMP perspective -- I agree 100%, you should know how your CPM is being sliced and allocated. But for other ad tech (like optimization or statistical algorithm models as examples) transparency may never be a reality.

Next story loading loading..