Programmatic ad buying declined in the first quarter of this year, which comes as no surprise after its reputation was blighted by issues of brand safety, fake news, and poor user experience, as well as an industry call to action for a return to quality over quantity.

But it’s all too easy to point the finger at tech and machines that promise scale and efficiency in ad execution. Programmatic is just a mechanism after all – a technological process that performs a task based on instructions. While the technology must be constantly reviewed to ensure it is working efficiently, it is the industries surrounding it that need to drive its effectiveness.

Misconception 1: Safe Ad Placements Are Tech Responsibility

There is a common perception that tech alone is fully accountable for the context in which ads are placed, and the content they appear next to. This is true only if the tech is selected and used with great care. In reality, proper planning & targeting, delivery, and control are the key elements to be managed.

If programmatic is used to deliver the right ad to the right user, based on user data, with little regard for the content, then the problem is in the planning and targeting. With programmatic, it is largely the profile of the user accessing the content that governs which ad is served, not the content itself. In this case, the technology is used to efficiently find the user rather than to decipher what content they are viewing.

Media owners, big brands, and agencies must take responsibility for fully understanding the digital context to ensure advertising is not supporting inappropriate content. Only by choosing advanced AI-based techniques that read the text as a human brain would and move away from primitive keyword filters, blacklists and whitelists will it be possible to reveal the true meaning of the content and determine if each impression is brand-safe before a programmatic bid is placed.

To make progress in cleaning up the ‘murky media supply chain’ surrounding programmatic, engagement, agencies and brands must look at the general attitudes and false perceptions that surround its processes. By utilizing AI and brand safety technologies, advertisers can still make use of the scale, efficiency and targeting precision of programmatic, whilst avoiding potentially damaging placements that may reflect badly on their brand or the industry in which they operate.  

Also Read: DMEXCO 2017 Roundup for Marketers: Marc Pritchard, Data Science, AI and “The Art of Narrative Disruption”

Misconception 2: Programmatic Alone Always Generates the Best Results

Programmatic is designed to optimize marketing performance, deliver correct ad placements, continually measure campaign success, and use insights to improve results against predetermined KPIs. But like all algorithmic processes, programmatic is only ever as accurate as the data used to drive it – namely its customer data. Simply buying multitudes of third-party data – which may be inaccurate, out of date, or biased – and expecting programmatic to come up with the right answers, is unfair and unrealistic.

To get the best out of programmatic, big brands must look to and activate their own first-party data to gain a deeper understanding of audience interests, how they consume content, and their propensity to interact with a particular message or offer. Using new techniques such as semantic analysis, behavioral analytics and propensity modeling, advertisers can enrich their own data and unearth valuable insights hidden within it.

By building these into actionable, 360-degree audience profiles, and adding engaging creativity and the selection of the context at the page level to the process, marketers can close the loop in generating interest and engagement with prospects and customers.

Misconception 3: Machines Are Smarter Than Humans

Even with huge advances in machine learning, we are still far from a time where technology is more intelligent than the human brain. Machines can undoubtedly be quicker and more efficient at performing a single task, for instance automated media buying, but they can only achieve this according to the rules, and within the parameters, set by human programmers.

Human review and intervention are still necessary to complement smart technologies in matters such as fake news. In highly subjective areas like fake news, machines with advanced linguistic capabilities can be the real-time barrier to filter controversial and extremist communication. However, in other elements, such as true fact and fiction checking and satirical communication, the human element is absolutely vital.

Even Google, which believes machine learning is more effective than people at identifying and removing extremist content on YouTube, is still using human reviewers alongside automated technologies.

What programmatic can do is inform the human process, reducing the workload of data analysis and campaign optimization, so the human workforce can focus its energies on more impactful areas such as the creative ad experience. Assisted Learning, in which machines are supported and guided by human team members is the true reality. 

Programmatic may have suffered a setback but the gains advertisers can make in scale, efficiency, and precision targeting are too great for them to abandon it altogether with almost four in every five US digital display dollars is still expected to be spent programmatically this year.

By attaining a better understanding of how automated media buying works, and gaining a more realistic view of what it needs to function as well as what it is capable of, advertisers can employ the additional technologies and techniques required to make the most of the programmatic process.

Recommended Read: Interview with Giovanni Strocchi, CEO, ADmantX