Why You Need Designer Data
One luxury of doing primary research is that you can get data perfectly tailored to the problem you need to solve. Primary research allows you to design how data will be elicited, and how, exactly, attitudes and behaviors are measured. Hence, you get the insight you need. If you don’t have perfectly tailored data, you need to be extra careful, and settle for less.
Here are examples of what I mean by needing designer data perfectly tailored to the problem you need to solve:
- If you are segmenting a market or developing personas, you need careful measurement (and comparable calibration) of each dimension hypothesized to be a differentiator. Quick inclusion of items via long lists on “select all” questions are not enough.
- If you want to report data for PR, or pull statistics into your sales and marketing materials for thought leadership, you need simple word-based measurement scales. Numeric scales or nuanced answer options will get lost in complications. You’ll find yourself struggling to report data truthfully.
- If you want to uncover drivers of outcomes, such as customer satisfaction, then you need large samples and measures that capture lots of variation. A simple Net Promoter Score (NPS) with an open-end probe of “why” will not give you key driver analysis.
Without these, you will pull your hair out, like we are right now with a project focused on driver analysis. The data come from an out-of-control survey that was written by a strategist who loves to say: “I’m not a quant, but I know enough to be dangerous.” Indeed, he is dangerous, because the quants cannot now rescue the survey, hit magic analysis buttons, and get useful results.
Here’s another example. We needed to know how many Americans had gone through a divorce within the past year, three years, five years, and ten years. It seems like those stats would be easy to calculate with so much good data available on marital status, including federally funded longitudinal studies on marriage and family. Nope. There is a ton of data out there to analyze, but none of it could be used to answer our question. Even for a seemingly simple question, we needed data designed for our purpose.
Remember this: data that is useful for market research is not just sitting out there, waiting to be analyzed with predictive analytics. That’s why the promise of being able to mine big data for deep insights has mostly failed our industry. Data needs to be generated by someone, by specifically measuring behaviors and asking people questions. Being able to design, generate, and tailor that data to your purpose makes a huge difference.
—Joe Hopper, Ph.D.