To simplify Programmatic Buying is an automated ad serving engine that can maintain multiple subsets of the campaign audience based on preset variables and show each subset of viewers a fully customized message or content, while allowing you to analyze and if needed change this content real time based on performance feedback. In plain english, programmatic can let you float thousands of campaign assets that would be customized heavily based on the user-like their location, time of the day, interests and spoken language. Also the system gives you data on how each type of ad is performing so you can make changes in a spreadheet for a particular category on the fly, making the ads more relevant as well as improve performance.
Which brings us to the challenges at hand; Programmatic Buying hinges around training an automated system to choose between messages and audience. This means the delivery must use relevent content and audience categories to begin with else the campaign will never pick up. Secondly no matter how well we try, this learning is an incremental one therefore requires time. its safe to say that the data mining and inferencing still has a long way to go but the current infrastructure can improve its own efficiency by filtering upto 90% of bad leads and clickthroughs. Lastly, the issue of viewability being balanced with bot clicks and fraudulent leakage with most networks in the range of 5% - 60% is a sizable task.
In the end Programmatic Buying is the obvious evolution of media- relevant and dynamic messaging that adapts to the business and the consumer on the go without a constant human workforce optimizing the campaign. Learn more by clicking the links below with some fantastic white papers by the Google Team.