Marketers understand that you simply
can’t build audience groups on pure demographic factors. After all,
Prince Charles of England and rocker Ozzy Osbourne are both British males of
the same approximate age. However, it’s safe to say that a marketing
message tailored for Ozzy wouldn’t necessarily convert the heir
apparent, Prince Charles. Consumer preferences, motivations, and needs
play a critical role in purchase decisions.
It’s clear that audience groups must be more sophisticated than
demographics. Even deep demographic factors like income or family status
don’t tell the full story. As Harvard Business Review(HBR)
highlights, the sorts of audience groups that convert are rarely 'created.' Instead, they’re 'uncovered' through data analysis that
incorporates behavioral clues from cookies, web analytics,
user-generated content, and other big data sources.
Why Your Audience Groups aren’t Converting
Despite the fact that marketers understand what’s required to build
audience groups, too few brands have segments that reflect reality. Information Week recently wrote about some of the 'perils' of big data analysis biases, which can include:
● Selection Bias
● Inclusion of Outliers
● Overfitting and Underfitting
● Confirmation Bias
The term 'data scientist' is ultimately accurate. To accurately
understand patterns in reality, marketing teams must leverage enormous
amounts of data to control against faulty results. If your big data
audience segments are based on false positives from too-small or
incomplete data sets, you could be suffering as a result. In one
anonymous case study detailed by Information Week, a brand’s profit margin decreased significantly as a result of audience groups’ creation that didn’t control for bias.
Do You Trust Your Audience Analysis Methods?
Many marketers have developed some level of big data fluency. They understand some common analysis methods used to develop audience groups, such as clustering or linear analysis. Undergraduate studies of statistics has leant familiarity with concepts like sample size and statistical significance. An abundance of easy-to-use analytics tools allows marketers without extensive technology backgrounds to perform complex analyses in a point-and-click environment. However, a lack of big data resources has forced many marketing teams to rely on pre-formed audience groups from 3rd party vendors that are questionable in accuracy.
One large-scale study by HBR
indicated that some 85% of product launches fail because of poor
segmentation methods. Ineffective segmentation can have a significant
impact on your brand’s profitability and outcomes. If you’re reliant on
pre-packaged audience groups that you’ve purchased from a 3rd-party
vendor, it’s likely time to refresh your segments. Join us as we review a
new approach to building audience groups that convert.
1. Form Segment Hypotheses
Big data analysis for the purpose of segmentation is inherently
scientific. The first step is to develop hypotheses about your segments.
Based on what you know about your segment, you can develop a framework
for analysis.
To avoid the risk of confirmation bias, your hypothesis should be based on known variables and goals. It could resemble the following statement:
'Individuals who are seeking a mortgage for a second home are often 30-50 years
old with an income of $100,000 or more per annum.'
A correctly-formed hypothesis serves to narrow your analysis, while still providing room to discover behavioral and motivational insights.
2. Obtain and Combine Data
By participating in a data exchange platform, marketers can gain immediate access to billions of data points in real-time. Marketers have the ability to set their own budget, and access insights on web behavior, preferences, and transaction history on consumers that match their existing contacts. Depending on your campaign goals and objectives, you can also opt to obtain contact information for additional prospects that match your goals and objectives. By connecting a data marketplace with your data management platform (DMP) tool, you can gain immediate access to fresh data insights.
3. Analyze
Effective marketing segmentation today has little resemblance to the mass marketing messages of yesterday. By obtaining third-party insights, you can gain a comprehensive understanding of how your contacts behave. This can lead to an understanding that your buyers prefer self-guided product research, are likely to have two children, or other rich factors that reveal segmentation without bias.
By allowing big data to form your segments without bias, you can avoid the risk of inaccurate results.
4. Launch Advertising
Once you have developed rich, up-to-date and accurate market segments, you can launch advertising to connect with your audience groups. Instead of relying on months-old segments created by a third-party vendor, your marketing team has the power to continually test, iterate, and improve your audience groups.