Become a stickler for the details in your research data
The devil is in the details! Our Group Client Director, Lena Cox, talks about the lessons she has learned about the importance of data protocol from working with big global brands like Coca-Cola.
There are times when you want someone to paint with broad, messy, rainbow strokes. This kind of unbridled creativity has a critical role to play in market research. Thinking outside the box, curiosity and passion are all essential qualities. But equally, we need a stickler with a check-list — someone with an eye for detail, who knows the rules and enforces them.
When it comes to data, we always want a strict protocol to ensure data quality. Because let's face it, it doesn't matter how great your analysis is—or how fancy the visualizations you share are—if the core underlying data isn't reliable and accurately reflect the market. At Infotools, we implement protocol and data quality for many of our clients and believe this step should be a "non-negotiable" for successful market research outputs. For clients who participate in things like global tracking programs across multiple markets, this approach is increasingly crucial for consistency.
Over the last three decades, I've learned many lessons about how changes to survey methodology can impact data quality and the ability to uncover market trends. These lessons ring especially true when working with clients who own multiple brands, operate across borders, and have many stakeholders along the research continuum. For example, Infotools works with The Coca-Cola Company, which owns more than 500 brands in more than 200 countries. Here are some tips based on our work with Coca-Cola and other clients:
1. Involve data expertise in the early stages of complex research design implementation
Clever survey design is even more difficult and complicated when you're aiming for maximum participant engagement by ensuring ease of use and minimal time spent. Survey design is a vital piece, but data processing is just as necessary. Failing to account for complex scripts in the final processes can mean that the collected data doesn't translate into the final data output correctly. The impact of this can be substantial.
To illustrate this, imagine building a pyramid representing brand equity (where the goal is to identify which brands have the strongest affinity with consumers – i.e., are the favorite/preferred in their category). For a brand with multiple sub-brands—let's say Grape Co. also owns Red Grape Co., Green Grape Co., and Vitamin Co.—it's easy to miss a step in your survey coding. Let me explain: If Question One is, "What is your favorite fruit brand?" and the survey participant answers, "Grape," your coding will list 'Grape Co.' as their favorite brand. However, what if the follow-up question is, "What type of Grape were you thinking of?" and they answer, "Vitamin Grape"? Have you coded the survey data to make sure 'Vitamin Grape Co.' is now listed as their favorite brand? It sounds like a small detail, but when you look at the big picture, failing to make this distinction could result in the conclusion that 'Grape Co.' is the favorite brand, when actually 'Vitamin Grape Co.' could be more accurate.
2. Minimize brand list revisions
Managing the demands of survey-data users can be a big job for many clients. If a company has multiple brands in the market, stakeholders can often disagree about the importance of one brand or another. Long brand lists impact the quality of the data collected and the experience for the survey participant, who may feel overwhelmed if asked about too many different brands at once.
For example, if a company like Coca-Cola is conducting global research, they have a high number of brands that they want to explore. From its flagship cola to Disani water, Powerade, Minute Maid, and so many more, it's easy to see how lists can get overwhelming. Add this to naming challenges across borders—think Diet Coke and Coca-Cola Light—and it becomes clear that parameters need to be set up ahead of time.
We can start to overcome these kinds of challenges by following these recommendations:
- Create and follow strict criteria of both when a brand can be removed from a study OR is considered important enough to be included in a study.
- Follow strict quotas on the maximum number of brands that can be included in any particular survey.
- Minimize the number of times the survey can be changed to increase the chance of identifying trend breaks. Where possible, calendarize brand list reviews to a couple of times a year to match up with periods when information is required for brand planning.
3. Be wary of the impact of logo design
Logo design and sizing changes can majorly impact survey data collection. Sometimes, even well-known brands plummet in claimed usage following what was considered a simple change to the logo.
For example, when one of Infotools' client's brands dropped drastically in the charts, we took a look at the logo within the survey. On review, the brand font size was so small, it was almost unreadable. It's no wonder the survey participants did not recognize it!
When using logos, it's critical to ensure no one image is causing bias to others within the survey. Make sure they're all set to the same size so that none dominate and that all are easy to read and recognize.
Another surprisingly common issue is discovering a mismatch between the logo used by the market research team compared to the logo, which is used in the market. You'd be surprised how often there is a disconnect between the marketing team and the market research team. So check what's actually out in the market and ensure the logo in your survey reflects what consumers are seeing.
4. Good sampling design and weighting is crucial
Some markets say, "We don't need to weight the data because we're applying strict quotas to our data collection."
But with a bit of digging, we often discover that there actually was some flexibility around quotas. Although small, these cause minor fluctuations in the consistency of the sample over time. Even if the weights only have a minor impact due to the quotas used, they should still be applied. Small fluctuations can impact the reliability of trend changes, multiplying over time to be much more significant, especially when overlaying population projection on the data. By weighting, we remove any risk the small fluctuations are impacting the insights.
5. Check, check, and recheck your script!
If your survey design is not implemented correctly in your script, then the impact can be very significant. The most common issue I see is the failure to implement brand rotation as specified in the design—which can be critical to even representation of the brands in the market.
The other key issue, which is becoming more important as increased complexity is included in survey design, is the correct inclusion of skips through the survey. If a skip isn't set up correctly early in a script, then the knock-on effect to later can be dire. So check, check and recheck the script against the questionnaire design to ensure all instructions have been included. And for global studies, you want a global script (i.e., all use the same tool), not just a globally shared document.
I know it sounds almost too basic, but always test the questionnaire from a survey participant's point of view. Put yourself in their shoes to see how they might react to what they see. Or better still, find a non-researcher to experience the survey and give feedback before it goes live.
As you can see, when it comes to data governance, you must be a stickler for the little things. This means being on the ground, knowing the ins and outs of survey data and market research. Sometimes it can mean challenging the status quo or putting your foot down to ensure the final results are robust. I believe it's critical for effective market research and makes all the difference to the final insights. So check in with your organization, who is the stickler with the checklist? The person you hate approaching when you're in a rush because you know they will force you to consider the little flaws in your plans? Go and talk to that person. Because now you know those small flaws can turn into costly mistakes down the line. (And if no one springs to mind, talk to the team at Infotools. We're handy with a checklist, and we know market research data inside out.)