Retailers benefit from behaviorally targeted sampling

What it means: “Big Data” has been one of the buzzwords of the retail industry for the better part of a decade, but so much of data’s importance is contingent on it being good, not just good enough. Quality data collection is increasingly more difficult to source, specifically in consumer data. The reality is that consumers no longer feel compelled to take surveys in the digital age, and that issue is even more pervasive among younger consumers. Consumer data often answers the why and how to support sales and financial transaction data, and finding solutions for consumer data contextualization is critical for retailers, market researchers and brands. Without the full 360-degree view of the consumer’s purchases beyond the sale, retailers could be left in the dark.

Consumers’ lives are busier than ever, and one consequence is that it is more challenging to identify willing, qualified and high-quality survey respondents for marketing research. Finding qualified respondents for research has been one of the biggest threats to not only the industry, but also to the retailers, who depend on this research to make critical decisions. Reaching the right sample respondents means higher quality research.

Behavioral targeting, also called behaviorally validated sampling, means that research participants have been selected at the outset based on a specific known behavior, typically purchase behavior. This starts at the sample recruitment process with a pool of potential respondents with verified qualifying purchase behavior, via “first-party” transaction data. These first-party data sources are usually:

• Shopper card scanned at point of purchase
• Online purchases with identifying information
• Self-scanned register receipts (relies on consumers to scan and can be less reliable and not complete)

Traditional gen pop sampling comparisons

Traditional general population aka “gen pop” sampling relies on screening panelists or potential respondents to identify who qualifies for research based on “claimed” behavior such as, “what did you have for lunch yesterday?”

For example, research may seek purchasers of a particular type of product or category. Some prior knowledge of potential participants, such as demographic profiles, that could potentially make qualification more likely, might be available. Nevertheless, for lower-incidence product categories like newly launched cereal, screening respondents from a general population pool can be time consuming and expensive.

Using traditional recruiting methods, finding shoppers who have purchased the right product is a little like finding a needle in a haystack, and it’s so much easier when we start out already knowing someone has bought the product we want to ask about, based on actual transaction data. 

More efficient research

84.51° looked at sample data from a range of recent consumer research projects conducted by 84.51° on behalf of Kroger and other CPG partners. 84.51° recommends that behaviorally targeted research respondents be “double verified” – meaning they must both qualify for the research based on: 

1. their historical purchase data, and
2. stated recall of the qualifying purchase in standard survey screening, to make sure that they were the purchaser for the household.

Because of this high standard, the research enjoys much higher incidence rates and is faster and more efficient to complete. We’ve seen survey incidence above 70%, even while targeting buyers of product categories with household penetration in the single digits.

Complex targets

Behavioral targeting can screen for complex behavioral profiles and segmentations that wouldn’t be feasible otherwise. Sometimes research targets are too complicated for even the most insightful consumers to be able to report their own qualifying behavior. Some examples:

• Lapsers who purchased a product in the past year, but not in the past three months
• New product triers who did and did not repeat purchase
• Brand switchers who used a coupon or responded to a price promotion
• Shoppers who did and did not conform to modeled predictions

As research respondents have become harder and harder to find, research industry panels have increased survey incentives, which has attracted less engaged and even fraudulent respondents. Survey takers have learned all the tricks: say ‘yes’ to as many questions as you can to be more likely to qualify, and then speed through surveys as fast as you can to earn as many incentives as possible. But behavioral targeting short circuits these strategies. We know a real person made an actual purchase with real money at a real store behind every targeted survey.

84.51° also found that when consumers are verified to have engaged in one behavior, they are less likely to claim to have engaged in others that can’t be verified, supporting the idea that some may be exaggerating the truth to qualify for surveys. Once in the survey, those who are not behaviorally targeted and verified are markedly more likely to speed through surveys impossibly fast and to “straight-line,” which is giving the same answer to every rating in a section.  

The study and data from 84.51° are consistent with the findings from a separate study by CASE4 Quality. This study showed that 40% of the traditional sample was unstable due to fraud (10%) and low engagement (30%). In fact, 10% of the surveys were completed by just 3% of the accounts and the average respondent enters 31.5 surveys per day. (See chart.)

Behavioral targeting brings this longitudinal lens to marketing research, opening up unprecedented opportunities for insight and providing predictive validity. Not only can we survey shoppers, already knowing the full breadth of their past shopping habits, we can follow them after they complete surveys, learning how and when they follow through and do what they told us they would do. 

Benefits of behaviorally targeted sampling  

Most importantly, all these differences can affect the bottom line. The priorities for brands, the areas of greatest strength and weakness and even which brand scores highest on different product benefits, can be different depending on whether the survey takers were behaviorally targeted or not. Real business decisions and actions taken by marketers could have been different.

It’s often hard to remember what we did yesterday, let alone what we purchased last month. Yet, much consumer research relies on consumers’ ability to recall their actual behavior. Behavioral targeting based on actual purchases provides multiple benefits including:

• Creates higher screening efficiencywhich makes research better and faster.
• Gives you access to new, in-depth complex behavioral profiles and segmentations that traditional sampling wouldn’t.
• Enables true longitudinal measurement to better track households and consumer behavior over time.
• Increases and improves research quality by ensuring brands are receiving the best research without fraudulent and unengaged respondents.

Using behavioral targeting in your research to uncover critical information about your brands provides a solid foundation for your business decisions. High quality research pays.

What’s next: As retailers and brands look to find solutions to the white space left by the declines in consumer survey responses, first-party data once again increases in value.  Retailers that are able to contextualize their first party data to unlock the “why” behind their consumer spend can harness the power of consumer data to make better strategic decisions. Behavioral targeting and other techniques enhance the quality of consumer insights, and that data can better inform decisions around retail media, assortment planning and other retail activities. Retailers who are utilizing data in this way will continue to win with consumers and brand partners, and all multinational chains need to invest in resources to support the growth in importance of first-party data and insights.