Analysis beyond market research

Suppose you are a manufacturer of a new type of cookware and you want to know what customers think about your products. Ideally, you should conduct a customer survey to effectively understand customers from a quantitative point of view. The responses to the survey questions will be digital, and the sample size of the survey population will be sufficient to draw conclusions. By surveying hundreds of customers, you will be guaranteed a valid statistical report.

For quantitative analysis, it is best to ask customers to rate the different performances of the kitchen utensils on a scale from 1 to 10, with 10 being the most important. Survey results may be: cooking speed, 8.2; food quality, 7.1; ease of cleaning, 6.5; kitchen space occupied: 4.8.

You can even study the correlation between these performance parameters and price sensitivity. You may find that price sensitivity decreases as the importance of cooking speed increases. In any case, the final results of the investigation will support decision-making. Your cookware might become a "quick cook cookware" and market it accordingly.

The traditional thinking of market research is implicit in this hypothetical scenario. Its characteristics are: information is the most basic, and quantitative information is especially favored. The larger the size of the survey sample, the more conducive it is to draw conclusions.

There is nothing inherently wrong with using market research methods to understand customers. Quantification is very efficient, and the more data and the larger the sample size, the better the results. But the problem is that the concept of market research is not enough to fully understand the customer.

Conventional thinking falls into the trap of simply assuming that as the amount of information available increases, so does customer understanding; The better the effect, the better the understanding of the customer.

Unfortunately, this limitation of data, especially sample size, is wrong. The understanding of the customer is not just some data, but above all an interpretation. To understand something is to parse how it happened, and part of the job is to predict what will happen in the future. But analysis is not just prediction, it is to explain and explain why and how things will happen in the future, for example, in the case of the new kitchen utensils mentioned earlier, what you are looking for is to analyze why and how customers use the new kitchen utensils.

Parsing does not come from data

In the search for parsing, it is important to realize that parsing never exists in data. Parsing is always separate from data. After reading the following case, you will understand this.

Let's say you work in the food service industry. You do a market research on a specific group of people and find that 85% of consumers in the survey group eat chicken at least 4 times during the week under investigation. Your survey of 400 people is not statistically representative. So you do another survey, select a sample of 2,000 people, and get the same results. In this way, you conclude that your survey is quantitatively accurate and statistically valid, and these consumers are eating chicken!

This case looks as easy as you might hope to understand the customer. Indeed, from a market research perspective, things are that simple. The question is: where is the parsing? You can predict that these consumers will continue to eat chicken (hence, you should add more chicken to the menu), but this is a prediction, not a parsing. You could say that parsing is simply extrapolation: if so many people eat chicken now, it's likely that many people will eat chicken in the future. But how do you know you can infer this way?

Sample statistics help to generalize the law of survey sample data to a larger population that should be surveyed but not surveyed. But statistics can't do anything about the difficult problem of making predictions from extrapolation from data. Survey results can indicate that people are eating chicken - or that they (the respondents) are eating chicken, but it cannot be inferred that this behavior will continue into the future, and although this inference is sometimes valid, it is only the investigator's Confidence is the result, not the result of logical thinking.

Looking at the issue from another angle, other interpretations are possible. Maybe the customers you surveyed chose to eat chicken to lose weight because they were concerned about their weight, or they were eating chicken to save money. Parsing in either direction, you can predict that customers will stop eating chicken when they have accomplished their goals or are fed up with their goals.

So your data - people eat a lot of chicken - can lead to two predictions: people will continue to eat chicken or they will stop eating chicken in the future. The same data can be applied to these two opposing forecasts. So, predictions only depend on the parsing, not the data.

So, where does the parsing come from? Logically, parsing cannot come from data, because parsing does not start with data. But if it doesn't come from data, where does it come from? Parsing comes from creativity. Parsing is the creative act of gaining insight into consumer behavior. This creativity may be motivated by data, grasping the state of affairs by observing and analyzing relational patterns in data, or by other creative impulses, including experience and intuition.

When a parsing comes to you, it doesn't matter where it came from, the parsing itself is fundamental. You put parsing first and data second.

Data can evaluate analytics

If you have an analytics in mind, data can play an important role in understanding your customers. The role of data is to make you more confident in the analysis you made, or to give it up and look for a better one. Data is the most powerful tool for evaluating analytics.

Let's go back to the new cookware case. First, you need to focus on parsing. If you have the data, it will give you inspiration to find insights for you. In this case, you might be thinking of an analysis where people value speed over quality. People are getting more and more hurried when it comes to food preparation, so they are willing to sacrifice food quality just to get it to the table faster.

You can then evaluate the parsing with data. First, does the consumer survey data you have at hand match this analysis? The results show a good deal: From a rating perspective, cooking speed is more important than food quality. This finding makes you more satisfied with the analysis.

However, despite the large sample size, the findings themselves should not give you too much confidence. The reason is that this finding would also be consistent with many other analyses that focus on cooking speed. For example, consumers may think that it is easy for a company to use technology to improve speed, but more difficult to improve quality. From this, the analysis will revolve around consumer expectations of technological possibilities and their real-world importance.

All parsing are different, they are divided according to two different dimensions. One of them can be called "special-universal", special analysis is only applicable to specific conditions, that is, this kind of analysis is only applicable to specific people in specific time and space. It is clear that this characteristic is more or less present.

Another dimension is "everyday-science". Routine analysis is presented from the perspective of the parties in the situation in which the analysis applies. This analysis may be an explanation of the client's own behavior, or the client cannot express himself clearly, so you can express it for him in a language he is familiar with. The analysis of kitchenware cases is expressed through people's own experience, so it is a daily analysis. The analysis of "satisfaction/sacrifice" in the case may be exactly what people may express themselves.

Scientific analysis is the opposite of everyday analysis, it does not represent real customer experience. This parsing is expressed in its own terms, which were invented in order to arrive at parsing.

For any kind of data, you need to form an analysis in your mind, which is either supported by the data or deviated from the data. In many cases, this parsing will be common sense. In fact, common-sense insights from focus group meetings and one-on-one interviews often contradict data from market research.

In the kitchenware case, the interpretation of "technical expectations" is easily refuted. For example, the following are typical responses from 10 "one-on-one interviews":

— "It always takes time to prepare food. Delicious mince pies just like Mom's."

— "There's no instant food or anything, I bet someone will come up with a simple and easy way to cook it, and there will be soon."

Just 10 interviews presented an analysis of "technical expectations"—that is, the target customer's belief that improving the cooking quality of food is less technically possible than increasing the speed of cooking. Is 10 interviews too few? Wouldn't 100 or 1,000 be better? From a sample selection perspective, yes.

But, keep in mind the broader issue, you have to go beyond the notion of market research. If your analysis is that people have special beliefs about technology, but if you see a class of people who should have special beliefs about technology and don't show that belief in everyday life, then that's enough to make a case for your analysis challenge.

Market research should be tied to numbers and needs to be more quantitative. But if you're dealing with a common-sense analysis that's highly exclusive and expressed in everyday language, then using data from focus group meetings and one-on-one interviews to make an analysis is less of a problem than It's easier to use data from market research. Many commonsense parsing and qualitative methods are a better match. In market research tests, these information methods are often too trivial to be ignored.

You need to go beyond market research to broaden your understanding of your customers. Understanding is a dynamic process, not a static pile of information. All that said, understanding is hard, you're trying to explain why people do things the way they do. Parsing anything is hard, and parsing human behavior is even harder. Parsing the behavior of acquaintances or people like you is hard enough, parsing the behavior of customers is even more challenging.

Excerpted from Kellogg on Marketing by Dawn Iacobucci, Understanding Consumers by Bobby J. Calder, John Wiley and Sons, Inc, with permission of publisher John Wiley and Sons, Inc. .Company registered copyright in 2001. Translated by Xiao Dongyan.

By Bobby J. Calder Charles H. Kellstadt Distinguished Professor of Marketing and Professor of Psychology at Northwestern University Kellogg School of Management for General Electric, Motorola, Aetna, Baxter, Bristol-Myers Advisor to Bristol-Myers Squibb, Prudential Financial Group, etc.