At its most basic level, marketing research is about getting information to help businesses make decisions and solve problems. That means asking the right questions, of the right people, in a way, they will be comfortable with and want to respond. Unfortunately, all three of these areas are rife with challenges, trade-offs, and opportunities. There is no one right way to do anything in marketing research.
However, the challenge of getting responses from the right people (a.k.a., data collection) is fundamental to data quality. While you may have a robust list of customer contacts, there are several reasons you might choose to use a non-customer sample. You may need a more general view of the market, you may be interested in speaking to individuals who did not purchase, or you may have concerns about over-surveying your customers. In those cases, you will collect your marketing research data from a purchased sample, either a commercially available list or a sample drawn from a panel.
There are four key challenges that arise in any marketing research project when planning for data collection:
- Who to talk to? You will want to talk to the person who has the information you need. For example, in B2B research, the end user of a product may not be the ultimate decision maker. So, if you are asking how and why a business decides to buy (or not), you will want to speak with the ultimate decision maker. If you want to better understand how a product is used, you will want to speak with the end user. Consumer marketing research presents similar challenges. Do you want to talk to the heavy user? Alternatively, the most recent purchaser? They may be the same person, or they may not be. Be careful in determining how you specify who will be your qualified respondent, as you could end up not getting the information you need. Additionally, sample definition has cost and methodology implications, as well. If you want to understand 12 to 15-year-old boys usage of hair products, you may have to pay more to get parent permission to survey kids under 18. Could it be that their Moms are the ones doing the purchasing? Could you get the information you need through a survey of Moms?
- How many do I need? “The more, the merrier” is not always true for marketing research, especially when it comes to purchasing a sample. When you buy a sample from a list broker or an online panel, you pay for every completed survey. So, the answer to the question of sample size is always dependent on your budget. In addition to the budget, you need to think about accuracy and risk. How accurate do you need the results to be? To get results with a margin of error of plus or minus 5%, you need of total sample of 384 (rounded to 400). For results with a margin of error of plus or minus 10%, you would need a sample size of 100 and for more accuracy, say a margin of error of plus or minus 2%, you would need to complete 2400 surveys. You need to balance the level of accuracy needed against the additional sample cost to determine the correct sample size for your project.
- What sample source should I use? We are beyond the days when we worried about the bias introduced by sampling only people on the internet, as well more than 70% of the population uses the internet. Therefore, for most marketing research projects, an online panel sample will be the most efficient and cost-effective sample source. Unless you have a good reason for using a different sample source (such as a small geography, or a rare population, or a specific B2B designation), most online panels should be able to meet your needs. Plus they have the added benefit that the panel company has already handled the recruiting, double opt-ins, and security issues. Given that most people do not want to participate in marketing research (see section on non-response bias below), people who are willing to join an online panel may be less representative of the population as a whole. You can mitigate that somewhat by working with a sample provider who aggregates or blends multiple panel sources to create a unique sample.
- Is non-response a problem? Contrary to popular belief, not everyone is willing to participate in marketing research. So, all marketing research data carries some amount of non-response bias. The question is not if you have non-response, but rather if it is a problem. If possible, you should compare your respondent group to another data source (such as population or Census data.) If your sample is demographically similar to the population you want to represent, your non-response should not create a problem in analysis. If, however, you find that you have significantly over- or under-sampled key demographic groups, you might choose to weight your data to better reflect your population and to reduce any possible bias that might have been introduced into your results.
The right respondents are the foundation of getting the right information to meet your project objectives. While there is no perfect sample, you can do a lot to minimize the potential for bias by thinking carefully about these sample and data collection issues before launching the survey.
To learn about Asking the Right Question, check out our previous blog: Want Better Data? Follow These 6 Tips to Ask Better Questions
Need help with your survey, Contact Us!