“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by James Aitken, chief executive officer at The Exchange Lab.
With 75% of chief marketing officers [PDF] using customer analytics to mine data, the practice of data-driven marketing is the rule rather than the exception.
Programmatic delivers efficient media planning and buying at scale, while sophisticated algorithms process vast volumes of increasingly granular data. Programmatic also adjusts impression bids based on a variety of factors, such as channel, frequency and geolocation.
Humans, however, are still a valuable part of the equation. While technology is vital for gathering and processing data, it’s the experts that turn numbers into insights and determine the right application of data for optimal campaign performance. In addition to advanced algorithms, marketers now need individuals – more than ever –with superior analytical skills to maximize data outputs.
A strategy that incorporates optimization should commence with prospecting, which creates a strong starting point for any campaign. Once data has been generated it can be segmented into performance tiers so that bids can be adjusted accordingly. Strong analytical skills are necessary to interpret reporting and assess criteria, such as time of day and weather, along with topical news and events.
Human analysis ensures the buying strategy can be refined and provides a solid base for automated campaign optimization. Without it, brands could waste their marketing investment by optimizing against the wrong metrics. With programmatic buying anticipated to account for 60% to 75% of display advertising by 2017, this could prove to be an expensive mistake.
Algorithms: Inefficient Without Human Input
Machines are necessary to automate and optimize digital advertising at scale with billions of ad impressions served every week and tens of billions of opportunities processed each day – far more than the human brain could handle. However, since marketing algorithms predict human behavior, their output is only as efficient as the human input they receive.
Data can only be fully harnessed when insight and analysis is layered with automation, creating a virtuous circle of refinement that allows precise and effective targeting. The algorithm can do the heavy lifting – adopting the mechanical part of the process – but humans are key for intelligent analysis and decision-making. Analysts also consider issues that machines may not, including unforeseen real-life events, seasonality and fraud, as well as having a greater understanding of the product being promoted.
Human Perspective Needed For Nuances, Color
Machines make absolute, black-or-white decisions, while humans can distinguish the gray in between. Without sufficient human input, algorithms can be overly restrictive, focusing too narrowly on high-performing impressions.
For example, an airline might experience most bookings on a Sunday morning when flights are at their cheapest. If the majority of inventory has been served on social sites on a Sunday, an algorithm might optimize placements to social sites and exclude all other days of the week. However, human traders would understand the importance of ads served leading up to the campaign – generating brand awareness – before consumers convert on a Sunday to take advantage of the cheapest fares. Eliminating all inventory except social media placements on a Sunday would be detrimental to the long-term performance of a campaign and could be harmful to the brand, so a human perspective is required to ensure the algorithm does not restrict bidding too dramatically.
Marketers should continually assess, interpret and react to insights provided by data in addition to the sophisticated algorithms that process the huge volumes of data. By leaving it all to the machines, brands risk missing the subtle nuances of consumer behavior. Where there are machines at work, don’t underestimate the human touch.
Follow James Aitken (@aitken_james), The Exchange Lab (@exchangelab) and AdExchanger (@adexchanger) on Twitter.
