One of the latest trends in today’s digital marketing ecosystem involves the intersection of Big Data, marketing attribution and programmatic media buying. It’s a trend that provides a glimpse of a future where the media buying process is much more efficient, and produces results far superior to traditional buying methodologies.
The way this cyclical ecosystem works is that programmatic media buying platforms such as ad exchanges like DoubleClick Ad Exchange, adservers like DoubleClick & Atlas, agency trading desks like AOD, demand side platforms (DSPs) like MediaMath, and real time bidding platforms (RTBs) like Marin, which function as independent silos, all produce marketing performance data . This data typically has a degree of granularity down to the placement level – channel, interaction, publisher, keyword, size, creative, etc. It includes these details on all the impressions and clicks of both converting and non-converting consumers exposed to a brand’s marketing efforts over a given period of time. When this amount of detail is multiplied by tens of thousands to hundreds of millions of marketing touches per day, week or month – by whatever frequency it is being collected and passed along to the next step in the process – one quickly recognizes what Big Data really means.
Once this Big Data is collected, it’s sent via automated feeds into a marketing attribution system, where it’s normalized into a common format and shared set of key performance metrics, and then run through the attribution modeling process to identify the cross-channel, cross-campaign and cross-tactic influence. Out of this marketing attribution process, a re-calculated set of metrics is produced that takes into account those intra-ecosystem influences, reflecting the fractional credit for an organization’s marketing success, now spread across the media buys that truly contributed to that success.
Once those re-calculated “attributed” metrics are produced, they are sent back to the programmatic media buying platforms that produced the original performance metrics – using the same automated feeds – to inform the next cycle of media buying being done by each platform. So despite the fact that these platforms are still buying media as separate silos, the attributed metrics produced by marketing attribution helps the platforms buy media in a coordinated, holistic way.
For example, say the raw performance data coming out of a particular display buy shows that it produced 100 conversions at a $10 cost per action CPA, where $9 is considered the maximum acceptable CPA. But after running that data through the attribution process – in combination with data from all the other sources – it’s learned that the buy in question actually produced 400 conversions at $2.50 CPA. Once this recalculated set of metrics is sent back to the programmatic media buying platforms, the platform that made the buy used in this example will funnel more budget to that buy, instead of ceasing spending due to the incorrectly perceived $10 CPA.
One More Level of Big Data/Complexity
At the point in the process, when the marketing performance data is brought into the attribution solution, it can be compared to demographic and behavioral data provided by data-management platforms through matching on anonymous cookie data, thereby enabling the attribution systems not just to recalculate the influence of every media tactic on overall performance, but on the performance of specific audience segments that the brand has defined. As the cycle continues, the attributed metrics produced by this process are also sent back to the programmatic media buying platforms of origin to inform demographic-specific and/or behavioral-specific media buys in the same way that they did in the example above.
A brave new world of marketing? Human interaction in media buying eliminated altogether? Hardly. But introducing marketing attribution to the programmatic media-buying process is what many brave new marketers are already doing to best leverage the Big Data at their disposal.