iMedia: A Media PhD

My colleague Cory Treffiletti is fond of saying that one lately needs a PhD to buy digital media these days, and I am struck on to what extent there is a nugget of truth to that. Just listened to a fascinating and lengthy summary of the third party syndicated research auditing process conducted by MRC (Media Ratings Council.) It brought to the fore some of the most intriguing questions: why do we trust the data we trust.

Earlier in the week I had a discussion where a group of us debated to what extent a data source is incented to have the most accurate data possible. It seems like a no brainer until you stop and consider that data accuracy has a sliding scale of cost-benefit. A resource like Nielsen NetRatings or Comscore must be PERCEIVED to have accurate data or no one will buy it. But the extent to which we are willing to pay top dollar for greater accuracy.

Naturally, what we want is 100% accurate info, at the lowest possible cost. So there is a point at which our need for accuracy must jibe with the cost of providing a service and letting a data company make a little on top of that. It would, for example, be great to have a sample size of 100% of web users but totally impractical from an executional and cost perspective.

I have used data from ACNielsen and Comscore for years, and I find both companies credible but not 100% accurate. And I think both would agree to that assessment. But as we enter into a World in which Google Analytics, Quantcast, Alexa, and Compete, among others are providing info that is guiding decisions, we need to have faith that these sources are providing accurate info. Happily, the MRC focuses on ensuring data methodologies, but it is something to think about before you refocus or alter your campaign or website or whatnot based upon a data source you know very little about.

Many of the new sources of data brag about huge samples, but huge sample does not necessarily accurate make. Methodology is critical as well, as is the composition of those huge samples. Look, all these new data sources may be wonderfully accurate, but we need to ask questions before we trust them. And sadly, many in the industry don't know what questions to ask.

Thanks for reading, and don't forget to write.