We’re researchers, so it makes sense that we love data. Big data, small data, and all data in-between. But does our love run so deep we overlook the validity of all the data we use? Deloitte Insights brings up a good point, “When big data contains bad data, it can lead to big problems for organizations that use that data to build and strengthen relationships with consumers.”

In this world of faster, more automate decisions and algorithms, we have an opportunity to take steps to validate the data we and our clients use. Earlier we discussed the importance of quality and integrity in the data we source, and now we’ll explore how data can be used to validate data used for audience targeting and other types of modeled or inferred data. If we don’t properly validate our data, marketers are at risk for poor messaging, misguided targeting and other sub-optimal decisions that will ultimately decrease efficiency and profit. Love is blind, so consider these use-cases for data validation.

(1) USE FIRST-PARTY AS GROUND-TRUTH

Although validating basic profile data is important, there are many other types of inferred data and models that can (and should) be validated. For example, passively-collected geolocation data are often used to create profiles or infer behaviors and interests based on models that make certain assumptions. These models need to be based on an accurate view of reality; asking a sample of data subjects about their behaviors and interests relative to their observed geolocation data can help to create better models.

First-party, permission-based data provides the ‘ground truth’ that is needed for accurate model creation and measurement.

(2) PROGRAMMATIC GARBAGE IN, PROGRAMMATIC GARBAGE OUT

Programmatic advertising and addressable media is becoming a larger and larger share of ad budgets. Advertisers are using advanced CRM-based strategies, modelling via survey segments and scaling with DMPs, and other algorithms to increase ad targeting efficiency. But automation and higher expectations drives the importance of accuracy. It is less efficient to measure audience accuracy when in-market or post-campaign since media dollars may be wasted on poorly-performing audiences. Audience samples can be surveyed to assess accuracy before media dollars are spent. This can provide confidence in the accuracy of the audiences upfront and allow client to gain the full benefit of their programmatic ad investment.