Qualitative data comes in many shapes and sizes. When performing cluster analyses for marketing functions, the value of incorporating qual data along with traditional quant metrics is paramount.
Qualitative data are pieces of information that cannot be accurately represented by common numerical characteristics or methods. To a market researcher, “qual data” can be the most valuable, but often the hardest to acquire and analyze en masse. This data type differs from the more common quantitative data, which is classified as a piece of data with a numerical characteristic or classification.
For those not too familiar with market research and data types, I provide the following examples below:
- Question: Rate your satisfaction on a 1-7 scale below (1 = extremely dissatisfied, 7 = extremely satisfied).
- Answer: 1
- Question: Please tell us about your meal in the space below:
- Answer: “The pasta dish was too salty. Next time I want the chef to make it sweeter.”
While the brief example above is centered on a post-meal restaurant survey, it showcases the differences between the two types of data. While both types arrive at the same general idea, one type without the other only uncovers half of the diner’s true experience.
More advanced marketing techniques such as consumer segmentation and product groupings require advanced statistical tools known as “cluster analyses.” This type of analysis is an example of machine learning, a commonly heard phrase in today’s data-centric universe.
A cluster analysis is a statistical model that arranges data into groups with similarities that are significantly different than other groups which share their own unique sets of similarities and characteristics.
Traditionally, cluster analyses have included quantitative metrics with little inclusion of qualitative data. Only recently however, has there been a way to quantify this qualitative data. Using a technique known as metaphor elicitation (analyzed through Meta4 Insight), market researchers uncover insights by asking respondents to select an image, and then to answer a brief question describing how or why that image relates to the question being asked. This data is then analyzed with a text analytics tool, and is codified based on similar words or phrases used throughout the responses in the data set.
By employing this technique and incorporating the results into cluster analyses, the statistical model can account for a “whole brain analysis”. This whole brain analysis accounts for both implicit and explicit thoughts that are uncovered using the combination of traditional quantitative and qualitative metrics. Rather than asking respondents to answer with a set of pre-formed responses, the free response aspect and interpretation capabilities of the metaphor elicitation exercises allow for previously uncoverd thoughts to be conveyed. Incorporating qualitative data into cluster analyses strengthens the probability that the research team has captured enough viable, and emotionally relevant data to feed into the model.
Traditional quantitative metrics have the ability to tell a compelling story driven by insights and hard numbers, but when qual data is incorporated into the mix, trends and groupings are uncovered that wouldn’t have been otherwise predicted.
Welcome to the new-age of market research.
Garrett Meccariello is a brand strategist and market researcher based out of Boston. In his free time he can be found building the next great brand, exploring a new city, and eating a lot of cured meat and cheese.