The biggest problem with “big data” in the advertising ecosphere isn’t bigness, per se. It’s the attendant noise factor (in scientific parlance) that thwarts the most diligent efforts to divine meaningful direction from so much third-party data… and there is a lot of data out there. How do you know what data is useful without costly rounds of trial and error?
Basic descriptive data such as estimated age and income can be a weak proxy for the target consumer; however, combining even a small handful of online attributes to define a target market often results in low scale and high data costs. Generic data clusters that group a large number of households into individual data segments with names like “soccer moms” and “urban strivers” have big scale, but they were never designed for precision, targeted marketing. The result is usually a larger-scale campaign, but not enough signal in the data to create meaningful targeting efficiencies. Unfortunately, it is the marketer’s budget that literally pays for the inefficiencies.
Finding the right data signals for a given marketing opportunity is an analytical exercise that should be based upon empirical data. This is why it is important to leverage the marketers’ first-party data to cut through the din and find the right targeting signals from within the vast sea of big data.
Marketers’ own first-party data, when coupled via predictive analytics with thousands of offline, third-party data points about consumers, dampens the noise to enable the most high-quality prospects to emerge, with a minimum of wasted ad impressions. High-quality offline data is stable, reliable and verifiable. It is not based on opinion or inferred information. It represents real intelligence about real people.
Think about it: Marketers already know exactly who their best customers are. They have lots of information about which of their existing customers represent the most lifetime value. So why not target people who most closely resemble these existing, best customers?
Take the online higher education market as an example. Many people of all ages consider taking online courses to pursue an advanced degree or to reinvent themselves after being laid off. Plenty of them will use a search engine or visit a website and fill out a form seeking information about course offerings, but only a small fraction of these prospects will actually enroll. In other words, completing an online inquiry form about online higher education is not a great predictor of someone actually becoming a student. Some online higher-education marketers have learned that the shortest route to their best prospects begins with a deep data dive on their most successful current students. This exercise can be performed by an outside data specialist, one who has access to many different data sources and who can algorithmically define the right combination of traits of a successful, high-value prospective student. By combining the power of many data sets into a single audience model based on a marketer’s best customers, this process can identify millions of high-value “look-alike” prospects who are more likely to enroll and successfully complete their degree program.
This custom-built, look-alike audience, derived from the marketer’s own first-party data, can be reached using relatively inexpensive real-time bidding (RTB) media inventory. Only consumers who match the high-value custom profile are served display ads, greatly reducing wasted ad impressions.
The same techniques work across a variety of product and service categories to reach consumers with the highest propensity to transact. Moreover, marketers can reach these consumers at the top of the proverbial marketing funnel, before they’ve exhibited any in-market online behavior — and before their competitors reach them. For all of its strengths, online behavioral targeting is not very effective at differentiating between high-value prospects and lower-value prospects. Moreover, it only works on the limited set of prospects who are showing behavioral signals. What about the millions of high-quality prospects farther up the funnel? To focus advertising dollars on high-value prospects earlier in the consideration cycle, one needs to look beyond behavioral indicators.
To be sure, behavioral targeting and high-value profile audience targeting are very complementary. Profile audience targeting influences the right prospects earlier in their consideration cycle, and behavioral targeting helps seal the deal at the bottom of the funnel. The beauty of high-value audience targeting is that marketers can reach highly profitable prospects before the competition. In order for the bottom funnel to grow, you have to feed the top. Just make sure that you’re feeding the upper funnel with your highest-value prospects.
Published 4/18/2012 on Adotas.com