Doug Conely is Senior Director, Global Data & Targeting at Tribal Fusion. Here he responds to a recent PostView coulmn, and argues that instead of the focus on the post view window argument, we as an industry should be moving to a weighted or linear attribution model.

I’m on record with you in agreeing that fewer, well placed, larger ads with better rich media executions should be the way forward. It will take time but I think enough people are pushing the larger and better formats to drive that change. However, publishers, while agreeing in theory, are yet to be convinced that fewer ads with placements likely to have view-time is a better monetisation strategy in practice (and no one wants to be first to jump!). The quantitative case has not been made yet but I’m optimistic that that will come.

I’m also on record as saying that last-impression-post-view is the least-worst attribution method available today given current infrastructure and education but it’s certainly not the way forward. Here I think you do the industry a disservice: we’ve now spoken with enough smart, passionate people with integrity who understand the post view issue that we’re convinced this too will change. Eventually, enough people will move to more sophisticated/ fair/ scientific (pick as appropriate) non-manual models that those agencies and third parties with vested interest in the status quo will be found out and commercial reality will kick in.

Reading your piece I think that the post view window argument is a red herring. The question is how much did each advertising touchpoint contribute to a measurable advertiser event? If you want to get rid of last view cookie bombing then a good first step, possible today, would be to move to weighted or linear attribution. That would incentivise people to manage to frequency (and there are enough studies to show that frequency is valuable) and efficient audiences (via look-alike models) as well as last view retargeting segments.

Longer-term though this is a “big data” problem of collection and processing (and maybe standards?). The weightings applied in exposure to conversion paths (or any other event you value) should be calculated not made up by rule of thumb. Our view is that optimising to those weightings is the next generation algorithm (after optimising to audiences which was the advance on optimising on sites and placements). But we’re then into an issue of statistics: will you ever see enough of a given exposure event to know with any confidence that it is a relevant factor?

I like your vision of the future... but I wonder if anyone out there is crazy enough to build a business model with long-term direct publisher relationships on quality content to drive more effective creative formats while developing exclusive segments to deliver audience discovery and look-alike models? They could price to market and then revenue share back to publishers so that they share the upside (and downside?). Would that be the next generation model... the BEAN? (Balanced Ecosystem Ad Network)