How can programmatic evolve to meet the needs of modern marketers?
The “black box” approach to programmatic advertising may work for less-savvy advertisers or those who prefer a hands-off approach, but platforms need to evolve to meet the demands of more experienced advertisers and agencies. Programmatic ad platforms without robust customization options will become a commodity. Like the ad tech version of a PC microprocessor, without the co-branding support of Intel Inside®.
There are five key components to an effective programmatic ad platform from the perspective of the advertiser.
First, programmatic ad platforms must leverage historical data when available. Guiding the algorithm with clear learnings will dramatically shorten the “learning period” discussed in the opening post of this series. It’s up to the advertiser or agency to come to the table with data points that differentiate them from their peers, and the platform’s job to provide structured means (csv imports, APIs, custom fields) for these details to be incorporated into their proprietary algorithm. This will reduce the time it takes for programmatic campaigns to find their sweet spot. Minimizing the loss of time and money is in the best interest of both parties. Happy advertisers spend more than frustrated ones.
Second, the programmatic ad platform user interface should provide users with insights into what is and is not working. Since most systems are black boxes to ensure the brilliance of the algorithm can not be ripped off, advertisers end up relying on their account rep to communicate key insights into the “who” and “why.” This presents a missed opportunity for advertisers to transfer those learnings into other aspects of their business, and for programmatic platforms to highlight their value up the chain.
Third, platforms must allow for customizable attribution models that reflect each business’ unique offerings, customer behavior and campaign goals. Campaign goals must be set to provide true value to the business, whether that be monetization, engagement or retention. Fluffy goals yield fluffy results. How much should a view-through conversion be worth for your business? How long of a conversion window makes sense based on historical time to purchase trends? Should there be a value assigned to users following your social media accounts or subscribing to your newsletter?
Fourth, platforms should allow advertisers to set clear goals and spend limits. Advertisers should be able to set a lifetime campaign budget and daily spend caps during the learning period. Progress to goal should be clear, whether that be interaction rate, cost per lead or return on ad spend. If the goals are reached, it should be easy to increase budgets. Similarly, if goals are not met after a statistically significant amount of data has come in, the advertiser should be alerted so that spend can be dialed back and strategy recalibrated. Set it and forget it only works for Ron Popeil.
Fifth, programmatic algorithms should adjust ad serving based on user intent signals. If the user was clearly doing research, encourage a social media follow or newsletter signup. If their website behavior is similar to past purchasers or they added abandoned their cart, add them to a remarketing list. Programmatic campaigns are great ways to get new eyeballs to your site, but not always the best means of converting them into customers. Sometimes, you need a second, third or even fourth touchpoint. That’s where other paid and organic traffic campaigns come into play. Programmatic may get them to the store window, but search or direct traffic brings in the sale while remarketing and email keep them coming back.
For programmatic to truly evolve beyond new customer prospecting, they will need to incorporate technologies like neural networks, which have a human brain-like ability for complex pattern recognition. Much like the breakthroughs seen in image recognition and language translation, deep learning has the potential to resolve many of the limitations we’ve experienced with programmatic platforms to date. In effect, allowing ad platforms to provide insights that even the sharpest human minds will miss. The platforms that lead this charge will be the ones that speed up the learning curve by making it easy to incorporate human context and raw data into machine learning output.