Commentary
Data Judging Data
- by Dina Hilal , January 12, 2012
With increased adoption of online audience targeting, more marketers and agencies are looking for ways to verify that they are paying to reach the right audiences. Historically, the advertising industry has turned to traditional panel-based independent sources to provide some measurement of confidence for audience quality. But a one-judge audience validation method may fall short.
How do we empower marketers with more information about who their message is reaching? The answer lies in the power of collective audience judging.
Let’s first talk about why a one-judge approach is sub-optimal.
In a world where consumer interactions with brands become increasingly complex, it becomes exponentially challenging for companies to deliver a validated, comprehensive approach to audience verification. Consumers are buying offline, shopping and researching online, creating social engagements, watching TV on their computers.
All these audience data points are being collected by different data providers (Bizo, Datalogix, Lotame, Nielsen, Polk, iXi, BlueKai, etc.), which specialize in niche data offerings, from business to social. Panel-based companies (or any one company) are limited in data scope and in their access to the larger Internet population. Most rely on data modeling and significant extrapolation, which impacts the accuracy of the audience verification.
That is why I believe a collective approach to audience verification is the better answer. The idiom “wisdom of the crowd” is the perfect analogy. Shifting the focus from a single point of audience validation to a collective body of industry data experts is a more efficient approach to verifying that dollars are spent reaching the right audiences.
By leveraging the collective wisdom of a wide range of data providers, marketers can get a clear profile of
their campaign audience before they spend a dollar on media.
Industry denizens like Lotame CEO Andy Monfried agree that marketers need unfiltered access to data from a number of
providers in order to evaluate an audience properly.
The most important point is that each type of data provider brings a different expertise to the judging table. When pulled together, they provide an unbiased verification far more valuable than traditional one-judge biased methods.
Similarly, marketers don’t have to weigh the accuracy between various validation methods (from panel-based to online actions to offline data), because it takes the blend of information to arrive at one simple answer.
Just like a medical diagnosis, marketers should seek a second (and a third and a fourth) opinion before making a final decision on their audience selection.
To put this into perspective, let’s say I am a marketer who is running a campaign targeting an audience likely to buy software. I have built a custom audience segment that includes a variety of categories from multiple data providers and would like to know how the experts would validate this audience.
Rather than making a decision using one isolated quality metric from one provider, a collective approach to audience verification would generate the following, based on real-world feedback from a diverse group of data experts. My custom segment of software buyers are:
• More likely to be IT professionals (according to Bizo)
• High-income earners of $100k plus (according to Lotame)
• In-market to buy videogames (according to BlueKai Intent)
• Frequent purchasers of carbonated beverages (according to Datalogix)
• Hybrid car enthusiasts (according to Polk Automotive)
Rather than answering -- am I reaching the right audience? -- I get the answer to what is the make-up of the audience I’m reaching?
This output is far more valuable
to me as a marketer because I’m not only validating a particular audience, but learning a more comprehensive picture of my target audience for expansion and creative optimization.
This also represents a more transparent approach to audience validation; an open platform allows all data providers a voice in judging the quality of data segments. And it wouldn’t be
restricted. It could be applied to all areas of audience targeting, including demographic, intent, search, offline and social.
Verification promises to play a more pivotal role
in the way audiences are measured and analyzed by marketers and agencies looking for greater ROI on their online media buys. Nobody is the king of judges when it comes to audience qualification.
Be sure to work with a data provider that offers collective audience wisdom to the judging table, so you can make the most informed audience targeting decisions.
Dina - Thanks for your thoughtful article! You raise an important topic. I whole-heartedly agree that it is important for a marketer to view their target market from many different data perspectives. Very few audiences can be precisely qualified with just a few data points. This is one of the reasons that we founded TruSignal - to help advertisers search through many different sources of 3rd party data and find the right combination of signals that precisely define their ideal audience. The “right” audience for an advertiser typically combines many data points to form a more holistic view - including past purchase behavior, demographics, hobbies/interests, geography, financial, et cetera.
I think of the audience validation problem as 2 fold: accuracy of the data and appropriateness of the data. Accuracy is about fact checking. If a data provider states that a particular user is 28 years old - how do you really know? Is the age data coming from stable, verified sources? Or is the age derived or estimated based upon some algorithm? The risk here is “garbage in, garbage out”. Since TruSignal is an aggregator of many sources of 3rd party data, we have the benefit of cross-verifying many of our data points. We employ the “wisdom of the crowd” validation within our own offline servers to verify the accuracy of the data we use to build our clients’ custom audiences.
Appropriateness of the data is about discovering which combination of data elements are the most indicative of an advertisers’ ideal target audience. How important are age and income to defining a given target? What about past purchases and geography or dwelling type? Consumers have many shades of gray. It is important to incorporate many different data points simultaneously to correctly define a target audience. At TruSignal, we address this challenge by comparing an advertiser’s 1st party data against many sources of 3rd party data. This analysis yields an audience formula that mathematically determines the right combination and weightings of data points to precisely define the audience. We use this audience formula to create a custom audience segment for one specific advertiser (available through the BlueKai exchange, of course!). This 1:1 approach allows TruSignal to incorporate the power of many different databases, into a single, actionable advertiser-specific audience.
Relying on just a few data points or the wrong data (not strongly predictive) is sub-optimal. We have found that using the “wisdom of the crowd” to factor in data from many providers is a much more effective approach.
David Dowhan
President
TruSignal
www.tru-signal.com
I'm not sure this is a critical when a robust set of actual purchase data exists for the target you are seeking. For example, if you had millions of explicit software buyers behavior on file, your target of "likely to buy software" would be pretty easy. This may not exist for software, but it certainly exists for consumer packaged goods. If you target is "likely to buy chocolate chip cookies" Catalina can give you exactly that target group for online display advertising.