Page 106 - The EDIT | Q2 2017

106
Thoughtleader
outcomes not longer-term business ones. It sucks because it measures the means not the end of media investment. It sucks because entire days of people’s lives vanish into reporting. And most of all it sucks because it breaks the first fundamental of marketing at PHD... that for brands to grow, they have to reach light and non-users of the category, which short-term optimisation totally sucks at.
Short-term KPIs and optimisation ONLY work when they are planned holistically with a longer-term view on marketing and business outcomes.
But long-term measurement has hardly covered itself in glory either. For every balanced view like the IPA’s The Long and the Short of It, there are a dozen publicity merchants who shout angrily and make fun at the (albeit short-term) efforts to understand the effect of media and marketing. What do the long- termists offer as a solution? Econometric Modelling, in principle great but hugely costly and backward looking — it can only make predictions based on what has gone before it. Alternatively, SOM / SOV analysis can point towards longer-term category share, but that only works if you’re operating in clearly defined category areas — which, increasingly, brands aren’t.
There’s even a case to be made that as things speed up any intention to have a long-term view and handle on a brand’s situation will just not be possible — the pace of category and innovation- driven change will outpace the ability of the long- termists to keep up.
What we will need to establish instead is the discipline of the mid-view. This isn’t the combination of short and long-term measurement into one view or dashboard, they are both too flawed as starting points. We need new measurements and currencies that sit at the edge of short-term effects and also at the boundary of longer-term measurement. Better, smarter, measures of our marketing activities that capitalise on the emergence of new worlds rather than trying to mitigate them.
The question then, is what does this all mean for the future of PHD’s planning product? I’m glad you asked.
PART FOUR:
NEW EXPERTISE
We’ve seen how AI and robotics, mixed reality, voice-recognition and VPAs will transform not just huge categories like transportation, finance, professional services and retail, but will have profound influences on our industry too.
The reality is that — as with any industry disruption — some roles will be created, some will change
and others will no longer exist. This, as PHD has identified, is predestined and inevitable. As Kevin Kelly has observed, in the future our effectiveness will be evaluated based on how well we work with machines.
With that in mind, let’s explore some of the roles that we would expect to see in our business once the future of our planning product is here.
The Attribution-Modeller
This will be one of the first significant new roles to emerge as our future planning product emerges. We will use reinforcement learning to analyse two data sets that will sit together in a closed-system. The first set of information will be attribution data
— information on which media impressions led to various outcomes. This will be hugely complex as most impressions have an indirect effect — either they are not immediately responded on, or they are designed for longer-term perception change rather than immediate action. Modelling and understanding all this data will be a specialist and important task.
The second element of the role will be to access sophisticated demand platforms — so they will be able to use their attribution models to inform which impressions are then planned and invested in across all possible formats and channels.
Future ‘Attribution-Modelers’ will be hugely important to clients as they will be experts not just in digital attribution (we can do this now) but they will (1) expand our understanding of modelling beyond digital ‘direct-action’ impressions into other channels via sophisticated modelling and analysis, (2) evolve a holistic view on measurement
THE EDIT ISSUE 06 | Q2 2017