The Art of Attribution

Mary Meeker’s annual report on Internet Trends always includes a slide on “time spent in media” versus “proportion of spend”, and the “laggard" that is mobile ad spend in this context (see pic). This is knowingly an incomplete approach to attribution modelling, as ultimately time spent doesn’t entirely correlate to “sales made”. At the end of the day as metrics go this is the only one that matters. 

The context, receptiveness, timing, creative capabilities of each media type play their own critical parts in the sales and marketing funnel. A sum of the parts analysis is insufficient. If Jaguar sells a car and attributes the sales value to Google as the last referral point to the location of the Jaguar showroom and reallocates their ad budgets to the detriment of TV and Out-of-Home advertising “because the model says so" they will sell less cars going forward. Yet a search marketer will make that argument. Attribution is a complex quasi-science, and there is a requisite level of cynicism required to any "econometric” modelling of advertising performance. 

As a tech company who entered the media market, one serious critique is of the structural silos that promote particular channels as opposed to overall campaign effectiveness (which is what the end customer cares about). Our world is outdoor media. Attribution has been, and remains a major challenge. In the UK, Route, the standard for audience measurement has helped take some big steps forward.

The future of attribution may well lie in the use of aggregated mobile-location data, where demographic movements in the real world are mapped in real time and digital screens serve content to crowds based on real time inputs, but you need some heavy tech infrastructure to power that data at scale. 

In the short run the major benefit our technology provides is in automating the processes of buying and selling this media, freeing up smart people to present scientifically defensible reasons to spend their precious resources on advertising.

Somewhere in the region of $30bn of advertising accounts are up for review currently, an unprecedented amount. I’m sure the main question CFOs and CMOs are asking as part of these reviews is why do we do this? Automation tech frees up agencies to engage with answering this question; the best answers will be transparent, honest and not silo-driven.