Jeremy Fain is the CEO and Co-Founder of Cognitiv, the first neural network technology available for marketers.
Marketers invest a great deal of time, revenue and resources to face the increasingly unsolved challenges of today’s advertising landscape. In the past, due to the antiquated practices of traditional programmatic advertising, marketers spent excess dollars and wasted energy on reaching customers, often without success or confidence that success was real.
Now, by leveraging deep learning technology, marketing can use algorithms to best determine where their advertising dollars are being allocated. I lead an ad tech agency focused on neural network technology for marketing, and I have seen how these advances in technology have the power to optimize programmatic advertising. This new transformation in ad tech, or programmatic 3.0, uses deep learning intelligence and custom algorithms to optimize buying decisions in real time, ultimately increasing transparency, delivering measurable ROI and enabling superior performance at scale.
To understand programmatic 3.0, one must explore the evolution of programmatic from its inception. The first iteration, programmatic 1.0, set out to fix the existing inefficiencies of ad networks; in traditional media buying, digital ads are manually traded. With the introduction of programmatic, both advertisers and publishers automated media buying, no longer needing to manually manage their inventory and “waterfall” of ad networks, and advertisers could purchase ad placements quickly and at low cost. Most importantly, programmatic 1.0 intended to provide both parties with transparency into the buying and selling process by cutting out any middlemen.
Unfortunately, advertisers soon discovered that this system was ineffective in increasing transparency. In fact, programmatic 1.0 sometimes facilitated fraud while limiting advertisers’ ability to see and control where specifically they were advertising. On top of that, brands and publishers were frequently hit with hidden tech fees, which further impacted their ROI. This state of affairs was clearly not sustainable, so evolution to programmatic 2.0, the self-service platform, was created to give advertisers oversight into the ad buying process.
While self-service platforms allow buyers and traders to see exactly where their advertising spend is allocated, there are still several key challenges. First, self-service by its very name requires large teams of traders to manage the day-to-day. In addition, self-service programmatic fails to automate at scale, forcing advertisers to participate in many iterations of manual trial and error as they attempt to optimize their campaigns. The combination of these adversities makes self-service programmatic a tedious, inefficient process that fails to provide advertisers with scalable, consistent results.
Programmatic 3.0 is adaptive algorithmic advertising in action. Through advanced machine learning, better known as deep learning, that uses artificial neural networks to sift through large amounts of data in search of customer patterns in real time, this new standard of performance marketing aims to address all of the problems of programmatic 2.0. Advertisers can create a custom algorithm based on their first-party data that will automatically optimize media buying and ad placement, thus allowing human traders to be more strategic and effective. Training these algorithms on real human outcomes, such as purchases, allows the deep neural networks to automatically weed out fraudulent signals and avoid them.
In addition to allowing advertisers to use their teams in more strategic ways, adaptive algorithmic advertising enables businesses to use large amounts of data to make real-time decisions and reveal insights quickly. These algorithms enable each impression opportunity to be evaluated individually, without a set of bulk rules that try to bucket millions or billions of impressions into an arbitrary commonality. This then allows very discrete and dedicated analysis during and after the campaigns are run, and allows results to be built into real-time, actionable insights. In other words, there is nowhere to hide poor performance.
Manual optimization has been made even more complicated by the vast quantities of data, which increases decision-making needs and exacerbates the lack of consistency. With deep learning’s data processing abilities, custom algorithms can take thousands of inputs at once, determine the best possible course of action instantaneously, and evolve over time as they receive more information about customer response to their advertising and market conditions.
The first two iterations of programmatic were, respectively, too opaque and too inefficient. Programmatic 3.0 aims to solve those issues by automating media buying while targeting unified optimization based on whatever goals or KPIs advertisers wish to achieve, thus freeing up time to focus on more strategic and creative marketing initiatives. Adaptive algorithmic advertising allows advertisers to drive performance at scale and get consistent results, all while providing a high level of transparency.
The adoption of any new technology can result in growing pains. In the case of programmatic 3.0, there are several challenges that advertisers should be aware of.
Firstly, DSPs are not currently set up to enable the use of deep learning algorithms, so brands looking to make the transition will either have to build the plugs that will allow them to implement their algorithms or work with a partner with the means and expertise to do so. Secondly, advertisers need to be cognizant of the fact that these algorithms need time to learn, and they will occasionally make the wrong decisions; however, their accuracy will improve the longer they are allowed to run. Finally, a deep learning algorithm is only as good as the data it is trained on, so advertisers must ensure they have a large, clean amount of deterministic consumer data on hand to ensure that the algorithm will be able to produce the desired results.
Instead of requiring human traders to manually optimize ad buying strategies, programmatic 3.0 is capable of optimizing on its own, which means that brands and agencies will have to fundamentally rethink the structure of their marketing teams. On a more practical level, unless brands themselves have data science teams to guide them through the transition, they will need to either build one internally or find a partner capable of building and training deep learning algorithms on their behalf.
Brands that take the latter route should ensure that their own customer data and desired KPIs are used to build the algorithms, as that will ensure that the algorithm’s actions are more closely aligned to the brands’ goals and customer base.
Forbes Agency Council is an invitation-only community for executives in successful public relations, media strategy, creative and advertising agencies. Do I qualify?
Go to Source
Author: Jeremy Fain, Forbes Councils Member