In this continuation of my series of tips for success in programmatic media buying I address audience targeting in programmatic campaigns.
Targeting network buys and direct buys has been both challenging and straight forward. In these traditional buys, targeting is often the key to success and media planners had to put in a lot of research and thought to get it right. Then, once targeting was set, the campaign could simply run with it. Things are a bit different in programmatic buying.
There is a class of targeting that is not discretionary. By law or policy, some campaigns must be targeted away from certain age groups. Other campaigns must be targeted by language. Still other campaigns are targeted to avoid conflict with concurrent campaigns. When regulatory compliance, common sense, or campaign strategy dictate, targeting should be inviolable. All media buying partners and platforms should adhere to this class of targeting and should be able to guarantee that 100% of your campaign will be delivered within the defined target audience.
There is another class of targeting that is applied to improve performance. This is the targeting that media planners and strategists worked so hard to refine. Now the practice of audience targeting for performance should change for programmatic media buying. Not all programmatic partners or platforms will discover the right audience automatically, but when you are working with one that does, adjust your targeting accordingly. Targeting that is entered into the system by campaign managers is only a starting point for automated targeting systems, but they do need the guidance. The entered or initial targeting lets the system concentrate its learning among audiences where there is likely to be success. The automated system will do two things. It will identify sub segments of the target population that are not productive and it will learn to avoid them. It will also identify high value segments outside of the target boundaries. If the initial targeting is a hard constraint, then the system will not pursue these audiences and your campaign will miss that opportunity. For this reason, initial targeting should be entered as guidance, not as a hard constraint, and you need a partner or platform that supports this functionality.
Sometimes, the right target audience simply can’t be known in advance. This is often the case with new product launches or new types of promotions. These are the toughest campaigns to jump-start. In such cases, there are a couple of techniques you can try. A learning pixel may be set on your site in advance of the campaign. Learning pixels pass information about the audience currently visiting your site so that the audience targeting partner can learn and build a look-alike audience. Be warned that look-alike audiences are a good starting place, but they are only the second of five levels of audience learning. (Look for a future post on the five levels.) Another method of zeroing in on an elusive or unclassified audience is to ask them. Running a poll that asks the audience to self-declare their preference with regard to the campaign is a very reliable method of uncovering the exact attributes of those consumers who will engage with the campaign. A poll can be run in advance of the campaign much like a learning pixel. A poll can also be run during the campaign, or embedded in the ads.
As with all aspects of technically driven advertising, targeting is something that works best in a real partnership. Trust what you know, trust your partners, and work together.
Next: Delivering Programmatic Campaigns