What astonishes me the most about this industry is how tried and true techniques are constantly being rediscovered and branded as new. This has happened in the digital world where the traditional modeling techniques from direct marketing are now starting to be applied at scale.
"What’s past is prologue" is a phrase that certainly applies to the evolving world of targeted display advertising. I have been a direct marketing analytics professional for almost 20 years and have witnessed more changes in the past five years than I had in the previous fifteen.
The explosion of digital technology has created an undeniable shift in marketing budgets, with more dollars being spent on digital marketing, often at the expense of traditional direct marketing channels such as mail, print, and email (when did email become traditional direct marketing?). The ability to tie offline identities to online presences has enabled the proliferation of more targeted display advertising and search bid strategies.
However, the more things change, the more they seem to stay the same.
The initial explosion of programmatic buying of display focused on the marketer’s ability to deliver advertising across the open web and online publisher platforms with bid strategies based on inventory and data. What has changed is that a brand can now dictate more control over the "direct" marketing that takes place online. Predictive analytics is finally catching up to the digital advertising technology. Initially, "targeted display advertising" really meant using digital tools to target display ads to third-party online segments found in cookie pools. Currently, using the same sophisticated modeling approaches that dominated the offline world for years, digital marketers can now focus on building and targeting online audience segments that are more receptive to their specific products or services, based on their own first-party data. Instead of purchasing advertising focused on inventory (private exchange, remnant, etc.) or third-party data only (e.g., canned “Auto Insurance Intenders” segments), marketers can now build and upload their own customized audience segments to ad servers and DSPs.
Three things you can do immediately to create customized online audience targets to deliver digital marketing:
- use your first-party data!
- build predictive digital targeting models using first-, third-, and even second-party data
- evaluate third-party digital data as inputs in your models – not all digital audience source data are created equal
Using "traditional" but powerful analytic techniques based on your own data assets gives you much more precision to drive targeted messaging to prospects across the web that are more likely to buy. Moving beyond the basic purchasing of audience segments from various DSPs allows you to clarify the clutter of online targeting and increase your ROAS. Big data and the "digital data exhaust" provided by online prospects as they navigate the open web change the build approach, but predictive models — targeting customer "lookalikes," response propensity, or lifetime value — rely on the same statistical techniques and provide the same power they always have. Applying these "traditional" modeling techniques to digital and first-party data can enable companies within the insurance space to selectively place display advertising in front of online prospects who are most likely to quote life, P&C, or healthcare insurance products.
Our role as agency partners is to help brands create targeted, connected, and synthesized marketing programs. Our role as digital analytic professionals is to move beyond canned, “off the shelf” online audience segments toward a more customized, powerful approach based on predictive analytics, using your own data assets.
The ability to leverage traditional hard-core modeling and advanced segmentation techniques into digital marketing is finally here. The exciting future of digital “direct” marketing is thus looking much more comfortable to a "traditional" analytics professional like me, who has spent years perfecting the art of using data assets and predictive, not descriptive, analytics to drive strategy. The past is prologue.