An update on attribution modeling - little has changed in a year



Just over a year ago I posted about attribution modeling as follow up to an SMX panel in NY. The key messages from that piece were:
  • Attribution was coming, but was certainly not here yet in the eyes of most marketers
  • Attribution modeling has the potential to make you smarter and your campaigns more efficient
  • There were many barriers to getting started, especially technical ones
  • But if the problems can be overcome then the results are worth it
So I watched the video above with interest - Adam Lavelle (CSO @ iCrossing) interviewing Chuck Sharp (SVP Analytics iCrossing) on this very topic, and importantly where this is heading. Of particular interest is that the industry has not moved much in the last 12 months - we still have many marketers using the last click scenario, and subsequently making budgeting decisions in a semi-blind fashion.

Why?

Well it is certainly not because of a lack of effort being poured into breaking down the technical barriers discussed in my last piece, with agencies and specialist companies all managing to extract the information that is required and put it together in some meaningful way.

What we actually see is that the continuing barrier is organizational behavior, and an unwillingness to change. I see two reasons why this might be the case, the first being about the level of accuracy that is possible, and the expectation of what level is acceptable or valuable.

As an example, I took my daughters to Disneyland Paris a few weeks ago. They researched the park online, they chose the hotel, they even chose the night they wanted the 'Princess Dinner', and all on their computer at home. They told me everything they wanted on the phone and I booked it on my laptop. Disneyland saw a direct load visit come in and spend a chunk of money within a 10 minute transaction and have no correlation, technical or otherwise, to any previous ad exposure, search activity etc. And the same problem exists when we consider a multi-browsing research phase spread across a desktop and a mobile device.

In short, there is no 100% accurate answer.

And that's ok! It is a directional tool, not one that will give 42 as the answer to Life, the Universe and Everything. It is designed to offer insights into how your channels are working together and for you to then make better decisions as marketers. In this data driven world, the value of gut feel from marketers should not be forgotten.

I particularly like what Chuck has to say in the video about what attribution is, where it fits and ultimately how it can be moved forward so we are not expressing these same frustrations another year from now.

Chuck talks about starting attribution as an internal research project; something with a question and an answer, and hopefully something that is actionable and can be shown to have made a difference. Such a project has a good shot at establishing some trust within the business, laying the foundations for further work to be conducted.

Beyond that, Chuck then talks about a 3 to 5 year timescale to see big changes in attitude, and that seems realistic to me. If the organization is public then it's likely to be a little longer due to the sensitives of showing year over year comparisons, and the limited opportunity in those reports to explain that the counting model has changed.

Most importantly - get started. Last click is inaccurate and is misleading.

FTC should ban all those evil cookies - the spoof!

Most display creative is crap