Foreign and domestic companies looking to invest in China or expand their business face a high level of complexity that makes it difficult to predict future developments. It is surprising that many managers fail to perform solid market research before moving into China, or that they stop it once they have established their operations. Some lack resources or are constrained by short-term financial pressures, while others are unaware of foreign data sources or are simply insensitive to the different environment.
Better-educated managers follow business school lessons and delegate market research to an in-house specialist or outside research firm. Yet such a decision often results in failure in emerging markets. In this article, I discuss how to conduct primary strategic market research and to analyse the results to create high impact.
Market research is a strategic task and needs to be led or monitored by top management, especially at the design and analysis stages, which create most value. Many Chinese market research agencies lack expertise. This also applies to many international agencies, since they usually work with less qualified local partners and employ local employees with lower skills than in the West. At the very least, a manager needs to co-ordinate in-house and outside specialists and examine their competence and performance.
Result-focused research design
Problems start with the way the research is designed. Questionnaires often do not elicit information that can be used as input for strategies. Managers must define the problem very precisely and develop hypotheses that solve a specific problem.
For example, I advised a Chinese financial institution that wanted to expand its insurance business in different cities in China. Based on experience and preliminary interviews, I hypothesised that because of their revenue growth, cities such as Shanghai and Guangzhou would be the best places on which to spend additional resources. To develop more detailed hypotheses, I asked myself, 'What do I have to believe for this statement to be true?'
First, I must believe that disposable incomes will grow in these cities. This implies that more people will enter an income bracket in which they can afford insurance and that those who already passed a certain income threshold can spend more on insurance. Second, the total population should grow as well. Third, a relatively high proportion of disposable income will be spent on insurance. Fourth, the company possesses the capability to reach and capture these customers.
This kind of upfront thinking cannot be delegated to 'specialists'. By taking such an end product-focused approach, the company can choose the appropriate data collection method and gather information in a targeted way from a variety of sources instead of blindly commissioning a large-scale survey.
The problem statement and hypotheses together can be used to formulate specific questions that elicit the required information. For example, the company may analyse GDP and population growth projections in selected cities. It may conduct a survey or focus group to determine the intention to purchase insurance, quantitatively or qualitatively. Finally, it may test-mail first-time offers to potential customers and additional product offers to existing clients.
The research design always needs to be pilot tested. Interviews can be arranged to test the effectiveness of questionnaires. In China, in particular, I have often found problems with translations. For many English terms, dictionaries yield the wrong translation. Pilot interviews help to incorporate idiomatic expressions. They may also show that a broader definition of competition is required.
The research design must also be checked to ensure that it meets legal requirements. For example, a decree issued in 1999 based on China's Statistics Law required foreign companies to submit surveys to the National Bureau of Statistics for pre-approval and the subsequent results for post-approval. The bureau officially justified this requirement by saying it needed to supervise the quality of survey firms and to safeguard national security. At the same time, as often happens in China, the order was left ambiguous and contained modifications and exceptions, to maximise room for discretion.
Submitting results, derived at high cost, to a government body entails the risk that, through leaks, competitors learn about them as well. Therefore, some companies just ignore this legal requirement. They also ask questions by telephone – a dangerous grey zone that may provoke authorities. Besides, phone interviews may not fit a research design that stipulates a broad population of respondents, some of whom may not possess a phone.
In view of the dangers and complications, it is better to proceed legally and develop a strategy for dealing with regulatory ambiguity. For example, companies may formally ask the National Bureau of Statistics to waive post-control for 'non-controversial' subjects.
Sampling to scientific standards
Managerial input does not end with the research design. Many Chinese market research agencies send interviewers out onto the streets. Such methods of 'convenience sampling' frequently create invalid, unreliable and unrepresentative results. For example, if 10 percent of the convenience sample in Beijing intends to buy your product, you cannot infer that 10 percent of all Beijing citizens will give the same response.
Great attention must therefore be paid to the pool from which respondents are selected, and the methods used to choose them. The sampling frame needs to include all members of the population, not just those who frequent a certain street at a given time.
Since China and other emerging markets lack sampling frames that approach the quality of Western databases (such as electoral registers), creative approaches are often needed. For example, I once conducted a survey in Moscow about the personal insurance market. First, I numbered all of the city's streets. I then generated random numbers for the streets to be included in my sample, and then produced random street numbers and apartment numbers. This way, I generated 1,500 addresses.
This example also shows that respondents should be selected randomly, ensuring that each element in the population has an equal chance of being chosen. Picking people on the street is not random, since the interviewer may use subjective criteria, and cultural influences may distort the choice. In China, for example, younger interviewers may be reluctant to interview elderly people because of the respect accorded to seniority.
To choose focus groups, where a small number of respondents discuss questions with a moderator, researchers do not necessarily have to sample randomly, but they do need clear and objective selection criteria. I once audited a focus group that was organised as part of research undertaken for a Shenzhen-based insurance company. I found out that one participant was an insurance agent for the company. Naturally, he praised its products.
I also discovered that there are 'professional' focus group participants in China who earn their living by taking part in such research. They know how to please the moderator and thus do not represent the population as a whole. The time when the focus group is held is another factor that determines who can participate. These examples all show that managers have to be vigilant and cannot simply delegate research.
Excluding bias from research
In China, many people conduct interviews for the first time in their lives with respondents who have never participated in market research. This lack of expertise is a recipe for disaster, since interviewer bias (for example, causing false answers as a result of leading questions) and respondent bias (for example, causing overly positive answers as a result of cultural habits) may distort the results.
You need to ensure that interviewers are properly trained, tested and guided. Saving on quality will inevitably generate meaningless or, even worse, misleading results. Interviewers need to master the theoretical bases for conducting research. Trainers need to ensure that they understand the questionnaire properly and refrain from leading the respondents in 'desired' directions. Before being sent into the real world, they should get experience and pass the required hurdles in mock interviews. They also need clear instructions, which should be summarised in an interview guide.
In focus groups, the moderator needs to ensure that the discussion concentrates on the questions and that all respondents participate actively. Real pictures and mental images often reveal more than words. For example, asking respondents what kind of person they associate with the various companies in the insurance market yielded valuable insights into competitive positioning and image. For example, the state-owned People's Insurance Company of China was personified as an old woman, whereas the client was represented by a dynamic stockbroker. Respondents voiced reservations about AIG, associated with the image of a Westerner, because it departed China abruptly after the revolution.
You also need to be prepared to deal with ethical problems. In my market research, I sometimes deliberately sent several interviewers to the same respondent. When no interviewer mentioned this duplication, I knew that something was wrong. I also called some respondents to check whether they had been visited. In focus groups, it is a good idea to observe sessions. For example, during the research for the insurance company I sat with the researchers behind a mirror glass and observed the questions and pattern of interaction. In Guangzhou, the group at times simply sat chatting without any direction from the moderator.
To exclude respondent bias, the research setting needs to be controlled and standardised whenever possible. An interviewee in a survey should be encouraged to answer questions freely. Because of Chinese politeness and fear of authority, research is often biased by the desire of respondents to please the interviewer. They may say, for example, that they would certainly buy a product, especially when they find out who commissioned the market research. Focus groups can diminish the effect of such cultural bias, since respondents may speak more freely when they interact with peers and do not feel observed.
Data analysis to create impact
After data collection, the manager should ask his staff to check the data input for accuracy. I have seen unqualified or unfocused typists making many mistakes in keying in answers from questionnaires. A small sample of answers can be compared with the inputted results as a quality check.
Some researchers spend weeks aimlessly analysing the collected data and fail to come up with results that can be acted upon. Like research design, analysis needs to be directed by hypotheses. Managers must compare the results with the original assumptions. To focus their efforts, they should ask themselves during each stage of the analysis, ?If I had to close my books today, what would be my answer??
Apart from such problem-solving discipline, managers need to add value to the analysis. Some agencies only produce cross tables tabulating mean values of two variables, such as income in the rows and product demand in the columns. For example, one box may show that customers in a high-income bracket are more likely to purchase a luxury product. This is merely a descriptive technique, which does not control for many factors that may cause, for example, spurious relations. Instead, strategists have to proceed analytically, developing models for drivers of the phenomenon under investigation, such as key buying factors, and determine how they interact.
Thinking creatively about change
It is of paramount importance not just to analyse the status quo but also to think creatively about future changes, which can be exploited to create competitive advantage. For example, I once developed a base case, best case and worst case scenario for cement companies in the Philippines during the turbulent Asia crisis. Using research input on government policy (introduction of new tariffs), construction growth (planned government infrastructure projects, and construction of office buildings and private housing) and competitors' capacity additions, I could develop intelligent views on future price trends. Managing capacity and imports turned out to be the key controllable leverage point. I always use the decision criterion that projects should go ahead if the worst case is still acceptable. Scenarios can also be generated by using predicted changes in key buying factors to forecast detailed changes in demand for a specific product.
Managers need to appreciate system complexity and system dynamics, such as the interaction of variables at one moment of time or over time. Instead of relying blindly on research input and analyses, they should perform 'sanity checks' to ensure that the data make sense. It is not always necessary to engage in sophisticated analysis. Often, an approximate calculation, which may establish orders of magnitude, will give an indication of whether the results are directionally right.
At the same time, managers should not use the excuse that they are not a 'numbers person' to reject sophisticated statistical approaches, such as structural equation modelling, which analyses cause-and-effect relationships between multiple variables (such as bad debt on the one hand, and income and educational background on the other), while holding measurement error constant. Managers should take some time to master these tools, in order to drill deeper into data, understand more and add value to discussions.
It is particularly important to recognise the limits of market research data, especially when dealing with product categories that are new in the market. For example, when NutraSweet first contemplated selling its sweetener in China, it was blinded by the size of the Chinese market. It relied exclusively on market research data from SRG, the correspondent of Nielsen in China, which used professional surveys, focus groups and in-home studies (where products were delivered to 100 households in Shanghai to see how customers used them).
In Shanghai, Beijing and Guangzhou, seven out of 10 consumers ranked the product better than substitutes (usually white sugar). Almost all respondents stated that they would use the product weekly. Based on the responses from the research, NutraSweet projected long-term trial purchases of 7m packets (at Yn18 per 50-pack box generating revenues of Yn2.5m in 1994). It forecast short-term repeat sales of about 64m packets, generating sales of Yn23m. Based on these optimistic projections, it decided to move into the mass market.
However, NutraSweet ignored several warning signs, such as doubts about whether Chinese customers are driven by the same level of health consciousness as customers in the West. Faced with disappointing sales – partly because the product was up to 10 times more expensive than sugar – it finally pulled out of the mass market, where per capita consumption was tiny, and instead focused on the narrow health market segment, in particular on customers suffering from diabetes.
Any analysis is an act of sense-making. Where you look determines what you will find. You may force data into a mental straightjacket and overlook other opportunities. In a large country whose complexity is similar to that of Europe, it is important to take account of regional variations rather than thinking in terms of a single 'China market'. Once again, managers need to unlearn business school lessons, which imprint on them the globalisation model and the value of large-scale standardisation in homogenous or converging markets.
Prof. Dr. Kai-Alexander Schlevogt (PhD Oxford), an internationally renowned expert in strategic studies, is the founder and president of the Schlevogt Business School, the first such institution in Germany to focus on emerging markets, especially China. He became the first permanent foreign professor in China (at Peking University) and served as a senior faculty member at the Australian Graduate School of Management (UNSW). He often leads education programmes for top managers and government officials in China and overseas. Website: www.schlevogt.de. Email: email@example.com.