Do-It-Yourself Marketing Research

Most early-stage entrepreneurs simply don’t have the resources to hire professional research firms to conduct the research that’s necessary to assess their ideas. Even if they had the resources, they might not choose to spend them this way, since the research may well show the idea to be a non-starter. So, if you’ve never done this before, how should you proceed?

What does marketing research do?

Marketing research – or due diligence, in financiers’ terms – is the design, collection, analysis and reporting of research intended to gather data pertinent to a particular marketing challenge or situation.

Marketing research is intended to address carefully defined marketing problems or opportunities. Research carried out without carefully thought out objectives usually means time and money down the drain. Other portions of your research may address different objectives – identifying the CSFs in your industry, for example. Whether the research you need is marketing research or some other kind, however, the guiding principles are the same.

Let’s begin with a model of the research process that sets forth the many decisions that must be made to conduct effective and actionable research. The steps in the research process are shown in Exhibit 1.

As this shows, the research process is fraught with numerous opportunities for error. That’s why it’s so important for entrepreneurs to be well-informed and critical users of the information that results from marketing and other research studies. To this end, each of the steps in the research process will be addressed, in terms of the decisions that you, the researcher, will need to make.

Step 1: identify the managerial problem and establish research objectives

A good place to start is to ask what the managerial problem or question is that a proposed programme of research might address. For most entrepreneurs, in their initial enquiries about assessing an opportunity, there are numerous managerial questions to be answered. How large is the market? How fast is it likely to grow? What segments are most attractive? Is the industry attractive? Who are the key competitors, and what competitive advantages might they have and not have if we enter? What customer wants and needs are not currently well satisfied, for what groups of customers or consumers? How likely are consumers to buy the solution we propose to offer? How much might they be willing to pay?

Taking each of these managerial questions one at a time, and applying appropriate analytical frameworks to each of them – such as macro-trend analysis, Porter’s five forces, and so on – provides clear guidance for the kind of information the researcher needs to obtain. The result is a set of research objectives (e.g. to determine market size and growth rate, to assess supplier power in this industry, to determine your target customers’ likelihood to buy, and so on) that will drive the research.

Step 2: determine data sources and types of data required

This step is critical in determining the cost-effectiveness and timeliness of the research effort. There are two key questions that the researcher must answer at this stage: should I gather data from primary or secondary sources? Whichever of these types of data sources are called for, do I need qualitative or quantitative research to satisfy my research objectives, or both?

Primary data are data collected from individual research subjects – using observation, a survey, interviews or whatever – that are then gathered and interpreted for the particular research objective at hand. Secondary data already exist – on the Internet, in government documents, in the business press, in company files, or wherever. Someone has already done the primary data collection and placed the data where others can access them, whether easily or with difficulty, whether free or at some cost.

Which are better – primary or secondary data? If (and it’s an important if) a research objective can be met using secondary data, that’s usually the best course to follow. Why? Three reasons:

1) First, it’s usually quicker to find the data somewhere than to collect them from scratch. Imagine having to collect demographic data without the census.

2) Second, it’s usually less costly to simply find existing secondary data than to collect them as primary data all over again.

3) Third, secondary data are typically based on what people actually do, or how they behave. Surveys, a common form of primary data, are based on what people say. The two are not the same, as we shall see in Appendix 5, on forecasting.

For entrepreneurs, secondary data, if they are available, should answer several important research questions, such as those on market and industry attractiveness at the macro-level. To identify sources for the particular secondary data you will need, consult a business librarian. A resourceful one who knows where to look for what you need to satisfy the research objectives you specify can save you enormous amounts of time. To explore consumers’ willingness to buy the solution you propose to develop, primary data are often necessary.

Where secondary data are to be collected, the entrepreneur needs to decide whether qualitative data or quantitative data are required. Most secondary research studies require both qualitative (e.g. macro-trends) and quantitative (e.g. market size) data. Fortunately, both are usually easy to find.

If primary data are necessary, a decision must be made about whether to collect the data using qualitative or quantitative research approaches. Qualitative research usually involves small samples of subjects and produces information that is not easily quantifiable. The benefit of qualitative data is that they may yield deeper insights into consumer behaviour than are available from quantitative research. For this reason, qualitative research is often conducted first and then used to guide subsequent quantitative research. An important drawback of qualitative research, however, is that its generally small samples may not represent fairly the larger population. Most experienced marketing researchers would say, ‘Never generalize from qualitative research. Always follow up with a quantitative study to test the hunches developed in the qualitative study.’ Such statements presume, however, that adequate research resources are available to conduct additional studies. Often, and particularly in entrepreneurial settings, such is not the case, and decision-makers are forced to rely, albeit tenuously, on small-scale qualitative studies.

Quantitative research collects data that are amenable to statistical analysis, usually from large enough samples so that inferences may be drawn with some confidence from the sample to the population from which the subjects in the sample are drawn. The principal benefit of quantitative research lies in its measurement of a population’s attitudes towards or likely response to products or marketing programmes. Because of their larger sample sizes and quantitative metrics, greater confidence can be placed in quantitative studies, when conducted properly, using appropriate sampling procedures and statistical techniques.

In most quantitative research, questionnaires are used that enable the researcher to measure the subjects’ responses on quantitative scales. These scales allow the researcher to compare different product attributes, the responses of demographically different consumers, and other differences in order to better understand some crucial questions:

(i) What products or product attributes do your prospective customers prefer?
(ii) Which product attributes are most important?
(iii) How satisfied are the prospective customers with one product compared with others?
(iv) How likely are the prospective customers to buy at different price points?

Where statistically significant differences are found, you can be relatively certain that the differences uncovered in the research reflect those actually found in the population as a whole. Examples of several kinds of quantitative scales commonly used in such research are shown in Exhibit 2.

Step 3: design the research

Designing secondary research is a simple matter of finding sources of information sufficient to satisfy the research objectives, and to ensure that the sources are credible ones. For primary qualitative research, such as focus groups or interviews, detailed guides must be prepared for conducting the research to specify what questions are to be asked. For primary quantitative research, research design is the most technical and most difficult step in conducting the research. It’s a good place to get professional help if you can afford it. The key decisions to be made in primary research design are to determine the data collection method and prepare the research instrument, to determine how to contact the participants in the research, and to design the sampling plan.

Determine the data collection method and prepare the research instrument

There are several methods of collecting primary data, of which the most common are observation, survey and experiment. Observation is just that: observing subjects doing something relevant to the objectives of the research. Typically, a form is prepared on which the observer records what is being observed. Many Japanese companies favour the use of observation to better understand not only consumers but also salespeople and distribution channel members.

Surveys involve writing a questionnaire, which will include questions and either scaled answers (like those shown in Exhibit 2) or spaces for openended answers, all of which are intended to capture whatever the researcher wants to learn. Demographic information about the respondent is also usually requested to aid in market segmentation and market targeting decisions. Constructing survey questions and formats for the answers is more difficult than one might expect, and is beyond the scope of this appendix, but several sources cited herein can help bring you up to speed on these tasks. Any business school marketing research text will have a chapter on questionnaire design.

Experiments are studies in which the researcher manipulates one or more variables, such as price or product features, either within the context of a survey or in a laboratory or field setting, in order to measure the effect of the manipulated variable on the consumer’s response. One common use of experiments is to examine the consumer’s likelihood to buy a new product at different price points. Different respondents are given different prices for the product, and the researcher tests differences in consumers’ likelihood to buy as the price changes. This procedure entails less bias than asking consumers what they would be willing to pay for a product, the typical answer to which is ‘As little as possible’.

Determine the contact method

Once a data collection method is chosen, the researcher must decide how to contact those who will participate in the research. Common choices include face-to-face (perhaps in a shopping mall or a public place), mail, telephone, fax, email and the Internet. Exhibit 3 shows some of the trade-offs that influence the choices you must make among these methods. A significant problem with survey research is that those who choose not to participate when asked (‘We’re eating dinner now, and please don’t call back!’) may differ from those who do participate. This non-response bias may distort the results of the research. Response rate can also be a problem, since many who are asked to participate will not do so. Response rates for mail surveys sent to consumers generally run around 15–20 per cent. Far lower response rates can be expected in business-tobusiness settings, which is why qualitative research methods, which use smaller samples, are often used. The other types are better or worse, as shown in Exhibit 3. Thus, for a mail survey for example, five to six times the number of surveys as you hope to receive must be mailed.

Design the sampling plan

Selecting a sample of participants for observational, survey or experimental research requires three questions to be answered:

1 Who is the population (or universe) from which the sample of respondents will be drawn?

2 What sample size is required to provide a level of confidence in the results that is acceptable to the decision-maker who will use the results of the research?

3 By what method – probability sampling (also called random sampling) or non-probability sampling (such as convenience sampling) – will the sample be selected?

Let’s discuss each of these issues briefly. First, the population from which the sample is to be drawn must be specified clearly. Typically, this consists of the target market, defined in demographic or behavioural terms, although excluding current non-users might not be a good idea for an entrepreneur who hopes to expand the market.

Second, the sample must be large enough to provide confidence, in a statistical sense, that statistical data, such as mean responses to survey questions, are truly within some narrow enough range, sometimes called the margin of error. In general, the larger the sample size, the smaller the margin of error. Exhibit 4 provides rough approximations of the margin of sampling error associated with different sample sizes.

Third, the idea behind probability or random sampling is that every person in the population has an equal chance of being selected. If non-probability samples, such as convenience samples, are used, then the sample may be biased. As a practical matter, convenience samples are used quite often for marketing research, because true random samples are more difficult and costly to obtain. Arguably, the non-response problem makes almost all samples potentially biased in the same way. An astute user should always ask what the sample selection method was. If the method is not random, then the user should enquire in detail about how the sample was selected to look for any obvious source of bias that might distort the research results.

Step 4: collect the data

By now, the hardest parts of the research process are complete, but the most time-consuming parts have just begun. Unfortunately, the datacollection process contributes more to overall error than any other step in the process. In some cases, especially where entrepreneurs conduct marketing research themselves instead of contracting with a third party, these errors are magnified. There are several common kinds of errors in face-to-face or telephone surveys that entrepreneurs should guard against:

selection errors by the interviewer (i.e. selecting respondents who are not members of the specified population);

collector bias: this occurs when the person collecting the data – inadvertently perhaps, in their enthusiasm for the opportunity – biases the respondents, so they tell the researcher what they think he or she wants to hear;

interpretation and recording of answers: in their zeal to obtain research results that support the feasibility of their opportunity, entrepreneurs sometimes have difficulty in interpreting their data objectively; in the end, the only people they fool are themselves;

in surveys conducted by fax, email or over the Internet, an additional problem is that the researcher does not really know who actually replied to the survey.

The data collection effort for surveys like these can be substantial. To complete 100 surveys with randomly selected homes using random-digit dialling, several hundred phone numbers and more than 1000 calls will likely be required.

Step 5: analyze the data

When the data have been collected, the completed data forms must be processed to yield the information that the project was designed to collect. The forms must be checked to see that instructions were followed, that the data are complete, and that the data are logical and consistent within each respondent’s form. Typically, the data are then entered into computer files, percentages and averages are computed, and comparisons are made between different classes, categories and groups of respondents. Often, sophisticated statistical analyses are required. If you lack the skills to do these things, you may wish either to obtain professional help or to find some marketing research students at a nearby university to help you with this phase – or any of the phases, for that matter – of your research.

Step 6: report the results

This is where the rubber meets the road. If the research study began with clearly defined research objectives, then reporting the results simply returns to those objectives and reports what was found. Where research is carried out without clear objectives – as is sometimes the case, unfortunately – reporting can be difficult, as no clear conclusions may be available. Including a report of the results of a well-designed marketing research study in a business plan can be a source of credibility for the writer and a powerful differentiator against other business plans. Perhaps the most common shortcoming of business plans that are rejected summarily by funding sources is that they lack any marketing research to provide support for the conclusions they draw. Wishful thinking and optimistic hand-waving are not enough.

What users of marketing research should ask

The research process described in this appendix makes clear where many of the potential stumbling blocks lie in designing and carrying out marketing research. Whether you conduct the research yourself or whether you hire someone to do it for you, an informed and critical user of marketing research should ask the following questions to ensure themselves that the research is unbiased and the results may be relied upon. These questions should be posed before the implementation of the research and again before its completion:

1 What are the objectives of the research? Will the data to be collected meet those objectives?

2 Are the data sources appropriate? Are cheaper, faster, secondary data used where possible? Is qualitative research planned first to ensure that quantitative research, if any, is on target?

3 Are the planned qualitative and/or quantitative research approaches well suited to the objectives of the research? Qualitative research is better for deep insights into consumer behaviour, while quantitative research is better for measurement of a population’s attitudes and likely responses to products or marketing programmes. For most entrepreneurs, the first of these purposes is the more important.

4 Is the research designed well? Will questionnaire scales permit the measurement necessary to meet the research objectives? Are the questions on a survey or in an interview or focus group unbiased? (‘Isn’t this a great new product? Do you like it?’) Do the contact method and sampling plan entail any known bias? Is the sample size large enough to meet the research objectives?

5 Are the planned analyses appropriate? They should be specified before the research is conducted.