Global Market Research: Roles, Methods, and Challenges

Introduction to Global Market Research

Global market research is moving away from its roots as a discipline that was detached from the business decision-making process, and is now more actively engaged with decision-facilitation. This shift has required new methodological thinking: a "holistic" analytical approach that provides clients with a rounded view of what all their qualitative and quantitative marketing evidence is saying. The new approach also requires analytical frameworks that combine hard global market research data with prior management knowledge and intuition. This paper also explains how the evidence accessible to global market researchers is changing, as a lead-up to outlining the critical thinking skills and interpretive power required to master the new world of information.

This paper discusses and explores the roles and challenges of global market research — both qualitative and quantitative — and the Internet as a tool in this research task, for developing effective global marketing strategies. Global market research has always been an authoritative tool enabling companies to measure customers' needs, wants, desires, and satisfaction. Nevertheless, with regard to customer satisfaction, if the data are not serviceable, reliable, pertinent, and timely, a company wastes its energy and assets. In addition, a business must react promptly to the information, or the whole undertaking becomes an exercise in futility (Van De Ven, 2003).

In describing how "new" global market research differs from the "old" modus operandi, it is helpful to think of global market research as operating on the following four fronts. First, how the quality of each piece of evidence will be assessed — robustness (Taylor, 2000). Second, the extent to which the new incoming information will be assessed relative to relevant and related past evidence — context. Third, the techniques used to evaluate the meaning and significance of each item of data — evaluation. And fourth, the way in which the research findings are presented to the client — application.

It is possible to characterize old global market research as being represented by the inner shaded area of Figure 1.1, which illustrates how old global market research typically functioned on each of the four fronts above. Robustness: the emphasis, in the past, was on working with orthodox concepts such as "validity" — is the evidence measuring what we think we are measuring, and is it free from any systematic bias? (Colby, 2007) — and "reliability" — how likely is it that the data will hold good over time, and that we will be able to reproduce our results? Context: in the past, most global market researchers would get no further than checking their new incoming study against, at most, one past related research report. Evaluation: this would inevitably focus on examining one data set and involve the application of classic statistical tests (Walsh, 1995). Application: the study would conclude with a presentation of the research findings, possibly with some recommendations for action, but would not be closely related to the subsequent decision-making process (Van De Ven, 2003).

New global market research takes us into new territory (Gabor, 2001), summarized by the activities shown in the outer panel in Figure 1.1. Robustness: the emphasis today is on "compensating" for the imperfections in the varied data sources that global market researchers now draw upon. Context: the availability of marketing information systems usually means that new global market research evidence will be set in a much richer context than ever before. Evaluation: orthodox statistical analytical methods will be employed alongside frameworks aimed at factoring prior management knowledge and intuition into the data analysis process, involving the analysis of multiple, not just single, data sets. Application: new global market research goes beyond simply presenting research findings and making recommendations — global market researchers are now much more closely involved with decision-facilitation.

The two broad approaches to qualitative data collection also form the basis of the frameworks that inform the analytical process. These two frameworks are as follows (Hogg, 2001). Rational: this approach to qualitative research is close to the scientific or positivist model, and is a style that will use comparatively few projective or enabling techniques (Van De Ven, 2003). The emphasis is on interpreting the responses to fairly direct qualitative questioning with sensitivity and insight, but without recourse to any particular psychological model. Interpretivist: this school may elect to draw on specific psychological and/or analytical models in making sense of the evidence, including the following approaches.

Freudian models: some analysts may approach their analysis using the Freudian paradigm. The "pure" use of this psychoanalytic theory is comparatively rare in market research today (Yin, 2003), but the techniques derived from it — notably "projection" — are still widely used. Another Freudian framework is Transaction Analysis (TA), which has its origins in Freud's division of the personality into superego, ego, and id. In TA, personality is seen as containing three ego states: parent, adult, and child. The "parent" contains instructions, moral codes, and obligations; the "adult" ego state evaluates and acts on information; and the "child" is the home for playfulness and creativity (Colby, 2007).

Organizing and Analyzing Research Data

Neurolinguistic programming (NLP): other qualitative analysts may focus their analytical framework around the concept of neurolinguistic programming. This can be helpful in studies where the researcher needs to understand how people absorb and process communications and information. NLP helps us to understand that people see things from different positions: in the first position, people see things from their own point of view; in the second position, from the point of view of the other person in the transaction; and in the third position, from the perspective of a neutral party. NLP reminds us that people receive information through various sensory channels — visual, auditory, kinaesthetic, and digital (mathematical or reasoned thinking) — and may have a preferred mode (Taylor, 2000).

Ethnographic approach: this takes its cue from observing how people behave in natural settings, rather than placing the emphasis on collecting information in a research setting. The commercial ethnographic analyst will place considerable importance on fully understanding the marketing context in which an individual is enacting their behaviour and expressing their attitudes (Weick, 1979). The ethnographic school of data analysis has a long tradition in social research, and is now again popular with commercial market researchers, although the commercial application of ethnography will fall some way short of the total immersion, over long periods of time, demanded by social research.

For most holistic data analysts, the dominant analytical framework will be a fairly eclectic and pragmatic one — sensitive to each of the various analytical schools discussed above, reflecting the rational school through its rigorous interpretation of hard consumer data, while also showing awareness of the value of subjecting particular types of evidence to various interpretivist treatments. On any specific problem, the holistic data analyst will choose the most appropriate framework for thinking, rather than committing exclusively to any one school.

The main analytical process in understanding qualitative research is one of going back and forth between the overall picture under investigation and the details of the data — the idea of developing a rolling hypothesis that is constantly checked against the available data, concepts, and principles (Ruff, 2004). The holistic analyst, in looking at qualitative evidence, will start by immersing themselves in the data and noting any overarching "big thoughts," then circle through the data again, picking up small clues that help verify the first big idea while beginning to identify the next.

The best analysis strategy is to first identify the evidence that has the biggest impact on the key issues and decisions to be taken. The analyst begins by identifying the overall storyline and comparing and contrasting themes emerging from different sub-groups, selecting information for its relevance and making decisions about which information will be central to the final presentation (Weick, 1979).

At this stage, the analyst looks for inter-relationships, shapes, and patterns — or alternatively, discontinuities — in the qualitative evidence. They will also search for metaphors that may help illuminate the analysis, or locate the data within the wider context of available models. To achieve an understanding of the big inter-relationships and patterns, it is often helpful to prepare some form of "cognitive map" or other "visual display" (Yin, 2003). This technique is designed to graphically map the inter-relationships between attitudes, behaviour, and different individual characteristics. The approach could start by constructing such maps for individual respondent case studies, and build up into a composite account of how groups of respondents are responding (Sang, 2003).

At the same time as understanding the big picture — following the principle of the Hermeneutic Circle — the analyst will also be examining the significance of different detailed findings. Some qualitative researchers will simply "absorb" the detail through close study of respondents' comments. Others will use specific techniques, including various counting techniques and content analysis (Rioux, 2005).

The holistic data analyst will constantly be switching between sorting the detail of a point and generating schemes and conceptual frameworks that explain what is happening in overall terms. This process continues until the analyst feels they have arrived at a grounded "resolution" of what the data is saying. The exact way in which different qualitative researchers will physically process the various forms of qualitative evidence will vary dramatically (Ruff, 2004). Different analysts will use procedures that work best for them on a fitness-to-purpose basis. For example, analysts who are good at simultaneously absorbing the big picture and the detail will place less reliance on detailed annotation systems, whereas those who are stronger on seeing the whole but may struggle to remember detail may require more elaborate systems for capturing detailed "quotes." It is clearly inappropriate to prescribe tight rules about what constitutes good qualitative materials-handling practice, though it is helpful to review some of the options available (Gabor, 2001).

The various types of qualitative research raw material that may be used include: audio recordings of group discussions and depth interviews; video recordings of the same; transcripts of the above; notes made during focus groups or depth interviews; post-group or post-interview notes; and comments and observations made by clients or research colleagues attending a group.

Some analysts opt for a process of total immersion and reflection, supported by listening repeatedly to group discussion or depth interview recordings, usually making notes as they do so. Some prepare, after each depth interview and group discussion, a set of notes on the main issues that arose, using these as the basis of the analysis (Hogg, 2001). Others make use of detailed typed transcripts with various marginal and annotational mechanisms to record significant "big picture" and "micro-points" on each page (Folpmers, 2008). Some researchers — working individually, in pairs, or in teams — "brain dump" from memory their observations and recollections of participation in the qualitative research process. Others literally cut and paste sections of transcripts, sorting and arranging them under headings related to the key research objectives. Some place the emphasis on quantification, constructing data grids or matrices along the lines of content analysis to assemble concrete facts under different headings, and some may use statistical software to assist with their analysis.

The way qualitative material is displayed is an integral part of the analysis process itself. It is the way the analyst begins to explain the data to themselves. Yet the various display mechanisms — matrices, cognitive maps, and various tabular accounts — also form the basis for how the data will be presented to the end decision-maker. However, the matrices used during the working process of understanding the qualitative data may be too elaborate for the purposes of the final presentation (Yin, 2002).

In most market research studies, the way language is used is clearly important, but this usually falls short of requiring a full etymological and syntactical analysis. Commercial market researchers need to take into account the way individuals discuss brands and the language they deploy. For most commercial research, it is generally accepted that the analyst will judiciously edit verbatim comments to avoid unnecessary repetition and deviation, while also correcting words that have obviously been used in error by respondents. It is also common practice to link together different short extracts from a longer verbatim comment, with connecting dots to indicate where a gap in the respondent's statement has occurred.

It is usually helpful for the holistic data analyst to develop representations that attempt to conceptualize key issues, thereby better conveying the overall storyline while still capturing the colour of respondents' raw feelings through verbatim comments. The challenge is to convey — by way of a one-page summary — how individuals feel about a complex issue and how attitudes change as the implications of their initial position are explained to them (Sang, 2003).

At the heart of data reduction is the goal of reducing the absolute amount of numbers on the page. This could include rounding percentages to the nearest decimal place unless the whole study pivots on the first decimal place, which is rare in most market research studies. It could also include cutting the data mass by grouping rows or columns that are not differentiating between categories and re-presenting them as a combined group. Careful attention needs to be paid to adding together percentages from different-sized bases, which involves re-percentaging rather than simply adding percentages together. Similarly, care needs to be taken when combining percentages from questions where respondents have been invited to give multiple answers, ensuring that data is de-duplicated so that one respondent's multiple mentions are not double-counted (Rioux, 2005).

Various averages or summarizing measures may be used instead of the full distribution. However, it is often the distribution that tells the real story in market research, rather than the bland column average. It is therefore important not to go too far in presenting purely measures of location (mean, median, and mode) without giving the reader a sense of the overall pattern of the data. Selective use of data — a further element of data reduction — involves making decisions about omitting unnecessary data that does not support the main arguments, either by excluding it completely or by presenting it in a supplementary form such as an annex or appendix (Sang, 2003).

Reordering and restructuring columns and rows: an alternative to combining rows and columns is to consider ways of reordering them from how they initially appeared on the computer printout. As a general rule, the dominant subject — for example, the brand — should be the row, and the sub-group variation — for example, region — should be the column. It may also be appropriate to reorder columns and rows from numerical highs to lows, or vice versa, or in a sequence that better fits the topic under investigation and the emerging storyline.

Adding clarity via improved labelling: the aim is to instantly tell the reader what the table is saying. Re-labelling the overall title of a table so that it is impactful and self-explanatory is a useful starting point. Re-labelling the titles of individual columns and rows more in line with the emerging storyline can also be effective. It is critical to maintain total clarity in labelling the following three scenarios: (a) "don't know" — a positive response to a question; (b) "not answered" — the person did not give an answer but did not say they did not know; and (c) where the question was not asked at all because respondents were filtered out (Hogg, 2001).

Re-percentaging to highlight a point: it is helpful to examine the data and consider whether a particular point can be better made by re-percentaging on an alternative base. For example, a study on attitudes towards the National Health Service may benefit from removing individuals who have had no direct experience of the NHS over the last ten years, so as to sharpen the salience and impact of the points being raised (Yin, 2002).

Taking out unnecessary noise: the computer printout may contain technically precise data which, paradoxically, could introduce comprehension difficulties. Column and row percentages rounded by the computer may produce totals slightly more or less than 100%, and the bases for different sub-groups may not add to the total sample because some respondents did not answer a particular question. One approach is appropriate editing — making small adjustments to remove irregularities with a footnote explaining the principles adopted. The other option is to clearly define to the reader what has happened. What is unacceptable is letting this noise from the original computer tabulations remain in the vague hope that it will not affect the reader's comprehension.

Imaginative use of design devices: the analyst should consider various design devices that can be deployed in a consistent regime to improve clarity. These can include the use of white space to separate columns and rows, the use of bold and dotted lines to differentiate the total sample from sub-groups, different typefaces (for example, bold for question summaries and italics for "not answered"), tinted and coloured panels, and icons or symbols to draw attention to particularly critical data. It is also possible to introduce short verbal summaries of particularly complicated data to reduce the data mass.

Literature Review: Holistic Data Analysis Framework

Seven general principles highlight the holistic approach to the analysis of quantitative data. First, having the confidence to put yourself into the analysis and accommodate prior knowledge: the holistic analyst knows that the starting point for making sense of the world is often their own "working prejudice" (Folpmers, 2008), which must then be inspected against management prior knowledge and the hard consumer evidence. Second, think in broad, simple strokes: the holistic analyst will always seek to add simplicity and clarity in quickly searching out the central storyline, rather than teasing out confusing minutiae likely to throw the decision-maker off the scent (Whyte, 2003). Third, think in both holistic and orthodox data analysis modes: the holistic analyst will blend orthodox statistical-based ways of making sense of the data with the central tenets and principles of holistic analysis.

Fourth, do not let the process kill the big idea; continually re-inspect the data against the end business objectives: the holistic data analyst will always be thinking about the end game — what piece of evidence is most pertinent in supporting or rejecting a putative decision. Their analysis will always be shaped by continually revisiting the research brief or proposal to ensure that every ounce of analytical effort is focused toward facilitating informed, evidence-based judgements (Hogg, 2001). They will not be drawn into a time-consuming process that loses sight of the whole reason for conducting the research in the first place (Rioux, 2005).

There have been few attempts to synthesize what the global market research industry knows about the interpretation of data into a single resource that would serve as a basic introduction to the holistic interpretation of global market research data for newcomers. One reason for this is that providing a synthesis of best global market research theory and practice — the art and the science — requires making difficult decisions about which points, from the vast body of literature available, to draw upon (Van De Ven, 2003). It is a literature that incorporates statistics, psychology, sociology, anthropology, marketing, economics, geography, communications theory, and much more. Much of this literature is also difficult to access for the busy practitioner, locked as it is in important but sometimes obscure publications. A further difficulty is that everyday global market research analysis relies on a body of knowledge that exists mainly in the form of proprietary techniques housed within individual research agencies, which are understandably reluctant to place this knowledge in the public domain (Van De Ven, 2003).

Before outlining the "holistic" approach to the analysis of global market research data, a brief explanation of the term is warranted (Rioux, 2005). The best place to start comparing global market research methodology with that of the natural sciences is to look at their common subject of investigation: the world. The naïve view assumes that the world is rather like a room full of furniture — comprising clearly defined and discrete entities that sit waiting for us to observe them. The more time you spend researching the world, however, the more a complex picture emerges. It turns out to be not a space full of objects, but a dense, interconnected web of forces and facts, blending almost seamlessly into each other, which we have to prise apart in order to study. Discovering these connections is central to the scientific approach, and taking account of the number, type, relative strength, density, and unexpectedness of these critical connections is a key challenge confronting such an approach (McClure, 2003).

The connection of overwhelming importance for science is the causal connection. Each of us is the centre of a vast network of causal connections radiating out into the world (Sang, 2003). A scientific approach aims to analyse these causal chains and identify the causes that contribute to the phenomena being studied. The way science does this is to analyse, classify, and break down the phenomena it studies. A central objective of the natural sciences is to isolate the phenomena under investigation, so that valid and reliable observations can be made. If unknown factors continue to play a part in the method devised for isolating the phenomena, the method will prove either unreliable or invalid. An unreliable method produces different results because an unknown factor may or may not be present in subsequent attempts to reproduce the phenomenon. An invalid method, even if the phenomenon was always reproducible, means the scientist is not measuring what they think they are measuring. These erroneous factors, which disrupt scientific methods by rendering them unreliable and invalid, are called artefacts (Harry & Schroeder, 2006). The complex nature of the commercial world makes these artefacts difficult to isolate, meaning that a degree of uncertainty in understanding and subsequent predictions must be acknowledged.

There is a paradox in scientific method. To acquire new knowledge requires you to understand how we know. To know something is to believe that it is actually true, and to believe it for the right reasons. If you believe — but for the wrong reasons — your chance of being right again in a similar situation is very slight. For example, in 1991, one of the authors, knowing nothing about racing, bet on a horse called Seagram to win the Grand National simply because that year's race sponsor was Seagram. As luck would have it, Seagram won. Yet the author could not claim to have any "knowledge" about how to pick winners. This way of reasoning would probably never work again; his reasons for believing Seagram would win were fanciful, and in no way connected with the complex array of causal factors that actually contributed to the horse's victory. We might therefore call an awareness of the limitations of the available ways of knowing the very hallmark of adeptness in knowledge (Van De Ven, 2003).

Implicit in all the aspects of science discussed above is the notion of building theories. Theory can be defined as those accounts which scientists make, in words, pictures, and equations, to describe a phenomenon and the various factors or variables that have a causal effect on it. Theory faces in two directions: backwards, in accounting for observations already made; and forwards, when it makes predictions about likely future events or observations (Sang, 2003). The more observations a theory can account for, or the more accurate its predictions, the more powerful it is said to be. To generalize across a range of observations, theorists often use metaphors or pictures that can be developed to generate new hypotheses (Whyte, 2003). There are three basic ways in which theory and observation can be related: deduction, induction, or abduction.

The method of deduction appears to produce the most certain knowledge: if all the beans in the bag are known to be white, and we take some beans from the bag, we can be certain they will be white. But it does so only by starting with a great deal of prior knowledge and not moving far beyond it. With the deductive method, the ratio of prior knowledge to new knowledge is very high.

The method of induction creates new knowledge by generalizing from individual observations. In this case, the observation is that the beans under investigation are from the bag, and the beans are white — leading to the prediction that all the other beans in the bag will be white. This method provides less certain knowledge than deduction, since the next bean could be a different colour. The more white beans we observe being produced from the bag, the more certain we can be of our conclusion. Conversely, if a non-white bean is produced, the hypothesis will be disproved or will need to be modified (Colby, 2007).

Conclusion

A further distinctive feature of the holistic data analysis approach is its commitment to presenting evidence in a way that will constructively impact on the decision-making process. This is why we have strongly commended the idea of presenting research evidence in the form of a compelling narrative. Not only is this approach known to keep decision-making audiences' attention, but it facilitates the integration of more intuition-based observations, which can then be evaluated alongside the evidence (Harry & Schroeder, 2006). It is also an approach that segues into the ultimate purpose of conducting global market research in the first place, namely to improve the quality of information-based judgements and decision-making. Looking to the future, there is the challenge for the holistic data analyst to develop techniques that will help with the process of "testing the safety" (Whyte, 2003) of evidence-based decisions — a stepping stone towards eventually being able to closely link how different combinations of research evidence have led to different types of successful and less successful decisions, and in turn to the different business outcomes eventually achieved (McClure, 2003).