In his 1970 hit book “Future Shock”, Alvin Toffler argued that as humanity progresses, the stress of an ever-connected, rapidly evolving, “super-industrial” world creates social paralysis. His main point — that more technology will not necessarily make life easier — certainly holds up when it comes to marketing data, and particularly the effect of programmatic on the display ecosystem. Programmatic’s complexities can lead to a similar slowdown in the way we handle our data, our jobs and our marketing campaigns in general. Let’s break it down:
According to eMarketer’s latest study, US programmatic ad spend will reach about $10 billion in 2014—that’s about 45% of the total US display market. It’s safe to say that most display schedules today have a mix of programmatic buys (managed through DSPs), and more traditional display buys (managed through 3rd party ad servers).
Moreover, about 55% of programmatic ads are served on mobile, and close to 10% of programmatic spend is on video. The average marketer now needs up to five ad management platforms to handle a typical display schedule:
- An ad server to handle conversion tracking and de-duplication across the Display media plan;
- A video ad server to handle the traditional video part of the Display schedule;
- At least one DSP to handle programmatic buys;
- A mobile DSP in case the specifics of the mobile programmatic schedule overwhelm the DSP’s capabilities; and
- A video DSP, in case the video programmatic schedule exceeds the DSP’s capabilities
… and probably a DMP to enhance audience targeting
Five years ago, a typical display media schedule ad operations plan looked something like this:
A typical campaign manager would traffic a campaign into a central ad server and go to that central ad server to get basic statistics about the campaign’s performance.
By 2014, a typical media schedule looks closer to this:
The generic 3rd party ad server is still central to tracking impressions, clicks and conversions but new buying currencies have emerged for video, such as cost per completed view (CPCV). The new measurement requires that the completed view metric be extracted from the video ad server and then matched with the appropriate media buy in the generic ad server in order to attain a combined metric for accurate analysis.
Serving, targeting and cost metrics for DSPs are only present in the DSP’s ad management platform. These also need to be extracted and matched to the appropriate media buy in the generic ad server in order to build ROI metrics for the DSP placements.
This increases ad ops complexity twice over:
1) Different people traffic in different platforms as these have become generally too complex for one ad ops specialist to handle
2) In order to reconcile important performance metrics such as ROI and cost of sales, ad ops needs to preserve a link between all of the platforms used on the schedule.
On the reporting side, not only do analysts now need to extract reporting from each platform separately, but they also need to make sure to reconcile the reporting to ensure that the appropriate cost is in front of the appropriate placement in order to derive correct ROI metrics.
A complete view of the campaign’s performance can only be achieved if all of these pieces are matched together and each media buy has all the metrics needed to compute performance.
Technology and service providers involved in the ecosystem ought to create smoother, better integrations in order to work through the increasing frictions in the ecosystem – and to prevent Data Shock.
Unlike Alvin Toffler, I remain optimistic about Future Data Shock. Those who survive organizational and structural paralysis will gain foresight and the ability to rapidly adapt. In our industry, this means creating ecosystem integrations that address the complexities of the ad management platforms above, and build them into our basic designs rather than add them as afterthoughts.
Katrin Ribant is cofounder and Chief Solutions Officer at Datorama.