How often have you heard somebody say about marketing, “it’s not rocket science”? Most of the time it is true - the principles and practices are straightforward, if requiring some care and attention. Complexity has been increasing, however, especially in the digital realm and in the wake of Big Data.
So it is refreshing to be able to say of something, “it really is rocket science”. In this case, it is a digital ad trading system devised by NASA scientists (among others) which is helping to transform the way media buyers optimise campaigns and attribute outcomes to inputs. Chairing a discussion over dinner last week, hosted by Rocket Fuel, I was able to learn at first hand just how challenging and dynamic the world of digital advertising has become - and therefore how necessary smart new tools like theirs are becoming.
Take this calculation: you are planning an ad campaign via a digital trading platform which offers you 12 attributes to select from across 145,710 segments. The number of different possible combinations to choose from comes to 3.53 x 1027 (see footnote). Without artificial intelligence to automate such decisions, they would never get made. And that is just one set of media buys - planners are involved with hundreds of clients running multiple, overlapping campaigns. There is simply not enough human resource to cope.
If you have been hearing a lot about Big Data and have been sceptical as to what it might mean, the digital media landscape provides ample proof that the concept is real and needs to be addressed. Advertisers want the best bang for their buck, publishers want to develop inventory that sells. Run lots of data through the right machines and both should emerge winners.
Getting to that point does require some thinking about, however.
1) Remember the 60/30/10 rule
Most of your advertising budget needs to go into activity which you know will deliver - that is the 60 per cent. But you also need to run some elements which extend your reach or coverage into non-core areas and add “colour” to the campaign - that is where 30 per cent of spend should go. These may not have the same level of payback, but they often help to multiply the outcome of the main ads. What you should not neglect is testing - spending 10 per cent on new, unproven options is likely to find you the next big breakthrough in media that could become your future bankers. That is where adopting new technologies, like digital trading platforms, are currently being brought into play. (If you need some maths to accept this approach, it is the Golden Mean ratio called “Phi”, by the way.)
2) Protect your pixels
One concern that emerged over dinner was the sheer number of tracking elements which are being put into online ads. As one planner said, “the pixel is getting very crowded”. Advertisers and their media buyers want ever more information about performance and are deploying multiple first party and third party cookies to get it. That is becoming something of a technical challenge - and may come under even more pressure if the Albrecht proposals for the Data Protection Regulation get adopted. These target anything that can help to identify an individual, which is likely to place beacons, cookies, pixels and the rest under an entirely new regime. Agencies may have to become more selective and cautious about what they allow into their ads.
3) Don’t forget the human touch
Machines may be essential to cope with the sheer volume of data and decisions, but you still need planners to take an overview. Humans are very skilled at instinctive pattern recognition and the experience of media agencies is a valuable resource. Sense checking what is happening via your trading platform is vital, not least because it may help to avoid the over-emphasis of “perfect fit” solutions that can lead to a too-narrow profile of ad buys.
Sitting down with a group of highly-skilled, technical professionals is always eye-opening (and at times a real stretch). What Big Data in the media world is demonstrating is that what was once considered a “black art” - media buying - is turning into a science like many other aspects of marketing. It may still involve black boxes, but it does make the basis for decisions more robust since they reflect underlying mathematical principles.
FOOTNOTE: For those of you who want the detail under that big number calculation, it goes like this:
18 Age * 2 Gender * 16 HHI *43,000 geos (US postal codes) * 100 Lifestyles * 800 Interest * 42 Psychographics * 990 Past Purchases of certain products * 17 Age of children brackets * 100,000 Contextual categories * 720 Time of Day * 7 Days of Week * 5 Ad sizes =3,529,513,709,568,000,000,000,000,000 (3.53 *1E+27)