Some of the most essential stages of marketing research process are as follows:
The process of marketing research starts with the identification of a problem.
Simultaneously the company also identifies whether the required information can be collected by the internal staff, or an external MR agency needs to be commissioned.
There is a realization that a marketing problem requires information to help find its solution. The marketing department asks the marketing research team to carry out a research, or seeks the services of a marketing research agency. If it is agreed that the services of a marketing research agency is required, the marketing department briefs the agency about the problem it wants the agency to investigate.
It discusses the nature of the problem in detail, and delineates its research needs. It is imperative that the agency learns as much about the client’s business as it can before it agrees to meet the client. The agency talks to existing clients who are in the same industry as the prospective client, and carries out exploratory research, like a search in the business press and on the internet to know more about the client.
The marketing department and the research agency meet to discuss why the research is needed, what it will be used for, when it is needed and how much is to be paid for it. The marketing department explains its problem and its research objectives minutely and meticulously.
It is important that the marketing department keeps in mind that while it understands the nuances of its problem, the agency will understand only as much as it is told. For example, if the company’s problem is to attract customers to a new product, it should brief the agency about the target market for which the product has been designed, the benefits and features of the product, its advertising and sales strategy for the product and a complete profile of its competitors.
The corresponding research objectives would be to identify customers who can use the product and to identify the features of the product that appeal to them most. The marketing department also explains as to how detailed the research should be, and how long the agency can take to complete the research.
The marketing department ensures that the agency understands its requirements by encouraging it to ask questions, and at the end of the meeting they collaborate to write down their expectations from each other. It is important that the marketing department hands over its requirement in writing to the agency, so that the latter always keeps in mind what it has to achieve for its client, and not let the research project go astray and esoteric. The written document can also be used to settles disputes between them, if there are any, at any time during the research project.
The research brief should be clear about the marketing problem for which research needs to be done. Additionally, the company must also take into consideration diverse perspectives of at least two or three MR agencies to ascertain the accuracy of their research designs.
i. Terms like market, market share and competitors should be clearly defined for the purpose of the research.
ii. Some researchers in the MR agency may be specialists in a particular data gathering method like group discussion, and they may bend research problems to use their favourite methods. This may be expensive in addition to being inappropriate for the research objectives under consideration. The company must familiarize itself with the MR agency and people handling its research project.
iii. Allow the researcher to ask even naive questions. Clearing doubts about the problem to be examined or related background information can avoid costlier problems in the future.
iv. Brief two or three agencies to get diverse viewpoints on the research problem.
The agency formulates a research proposal, in which it delineates as to what it will achieve for its client and how it will go about doing so. For example, it may agree to identify demographic variables of the customers of the client’s new product by a certain date and at a certain price. This is almost always written so that both parties understand each other’s commitments.
A client is likely to be impressed by the research proposal if it demonstrates that the agency has understood the client’s problems and its research objectives clearly. A good research proposal unambiguously describes how the research will be conducted-What will be the sample size? Which survey methods would be used? How would data be analysed? How long will the research take? In what form would the report be presented? And how much will the company have to pay?
The client judges the intent and competence of the agency by what is contained in the research proposal. Therefore the agency should be forthright and avoid camouflaging of any type—it should simply state what it is going to achieve for the client, how, and by when.
The agency should ensure that the client understands the nuances of the proposal, and the client should ensure that the proposal contains everything that it wants the agency to do. The client should be aware of the fact that the agency will do what it has promised in the research proposal, but no more, so it should be exhaustive in listing its requirements, and ensure that the agency has promised to meets all its requirements in the research proposal. If there is any doubt about an issue, the client should ask the agency for clarification.
The research proposal should first define the problem. For doing this, the researcher must consider the purpose for which the study is being conducted, the available information, additional information required and how such information would aid the client in the decision making process. The problem can be formulated without much consultation, though frequently it may involve discussions with decision makers in the company, experts outside the client’s company, use of secondary data or even some exploratory research, for instance, a few focus groups.
This should be followed by developing an approach to the problem. This involves developing a framework for analysing the problem. Relevant questions, for which answers are needed by conducting the research, must be clearly propounded in this step.
The research design should explain the methods of collecting data, sampling plan and a plan for data analysis. The client should be clearly told about the sources from where the relevant data can be obtained, the time frame within which the entire study will be completed and the modus operandi of conducting the study.
Marketing research is primarily used to enable managers to take decisions. However, some limitations such as the inability to define the problem accurately, unavailability of past researches, etc., may force researchers to gather more background information about the problem in hand. In such cases, exploratory research is used first. Thereafter, conclusive research is done either through surveys or experiments, enabling managers to take decisions. Decision making is not possible through exploratory research.
The agency conducts exploratory research, which is a preliminary exploration of the research area prior to quantitative data collection stage. It is normally done after the client accepts the agency’s research proposal, but it can also take place before the client-agency briefing meeting and submission of the proposal. It can also be used in the problem definition process.
Hypothesis generated at this stage can be tested further by undertaking quantitative research. The idea is not to collect data and form conclusions but to get better understanding of the client’s target market, its customers and its competitors. So when the agency starts collecting data, it knows who its respondents are, what it has to know about them and what it has to ask from them.
During the exploratory research, the agency tries to identify and understand the respondents who would be interviewed in the data collection stage. It also tries to get a comprehensive understanding of the industry in which the client operates its target markets and its competitors.
Exploratory research helps in getting rid of the prejudices that the agency might possess about the research area. Such prejudices involuntarily misguide the research proceedings and are a major deterrence in understanding the realities of the research area.
Exploratory research is also often useful under such circumstances when enough information does not exist about the area being researched. Therefore, the agency conducts its exploratory research diligently. Exploratory research is used when the available data is insufficient for proceeding further. It enables the MR firm to gain better insights about the problem in hand.
i. Secondary research:
Secondary data is compiled by other people, for other purposes and is not meant specifically for the research in question. Secondary data is generally used to understand the problem, to define the problem better, to develop an appropriate approach to the problem, to formulate a suitable research design or to answer certain questions. It can also be used to interpret primary data more insightfully.
Internal records of the company are important sources of secondary data and so are reports of previous researches carried out by the company. Therefore, it is important that a company maintains data like revenue, profit and expenses of its products and major customers.
Government, trade associations, newspapers, magazines and internet are important external sources of secondary data. It is easy and rather inexpensive to obtain secondary data, therefore an agency should spend some time carrying out secondary research. It may also happen that an agency might get most of the data of a research project from secondary sources and may be saved from conducting primary research. But, if it does not conduct secondary research diligently, it may carry out primary research to obtain data that is already available through secondary sources.
ii. Qualitative research:
In a focus group: unstructured or semi-structured discussions are conducted between a representative group of consumers and a trained moderator, who is often a psychologist. The idea is to let consumers discuss their preferences, behaviour, motivations, attitudes and beliefs in an uninhibited manner, in an unthreatening environment. And it is possible that a company learns about its customers’ nuanced needs and behaviour, which they might not have been able to reveal in a survey.
The moderator leads the discussion and ensures that the group stays on course and discusses the issues that are on the agenda of the focus group. But the moderator understands that the group cannot be allowed to become inhibited, and therefore, allows the members to discuss issues which are important to them, but which are not on the agenda of the focus group.
Focus groups are valuable due to the unexpected findings that emanate from free-flowing discussions among customers. Such discussions are likely to reveal complex and subtle relationships between consumers and products. The discussions are not based on percentages, averages or other numbers that are impersonal in nature. And as marketers are directly in contact with their customers, focus groups often stimulate new ideas for them.
The resultant data from focus group discussions is interpreted by experts. The data obtained from discussions is helpful when designing the questionnaire for conducting quantitative research, which will focus on what is important to the respondent. The instrument is thus worded in the language that the respondent uses and understands.
However, focus groups are not representative of the population, and hence cannot be the sole basis for decision making. For instance, the size of the target audience holding a particular type of view cannot be gauged by conducting focus group discussions.
In a depth interview, a consumer is interviewed at length about an issue. As in a focus group, the idea is to understand customers’ nuanced needs, behaviour and motivations, but an agency resorts to the more expensive depth interview, when the issue is such that presence of other members inhibits honest answers and viewpoints.
Also, when an agency has to understand an individual’s response to a stimulus such as an advertisement, or has to understand his decision making process as when an executive buys an expensive equipment, it sanctions a depth interview.
Depth interviews are also used when an agency is not able to get a group together—for instance, when it is dealing with busy CEOs, it mandates a depth interview. The questionnaire is usually in the form of an outline, allowing respondents to freely express their views about the topics being questioned. Thus, there is a lot of flexibility for the researcher in this method, and responses that are obtained may yield new ideas which otherwise could not have been obtained by means of structured questionnaires.
Depth interviews are difficult to standardize as no two interviewers are likely to proceed in an absolutely similar manner, and their biases are also likely to reflect in the results if the interviewers are not adequately trained.
The result of the qualitative research has to be studied carefully as it is based on small sample sizes. Much depends on the ability to interpret subjective data and may vary from one analyst to the other. More interesting or surprising viewpoints may be disproportionately reported. The quality of the data obtained also depends on the skills of the interviewer. Qualitative methods of data collection take longer amount of time and are more expensive.
The agency consults experts, who may not belong to the company’s target market, but who are knowledgeable about the issues being researched, because they have expended considerable time and resources studying it. They are very insightful and possess insider knowledge about the markets in which they possess expertise. Consumers want to look good when they talk about themselves, therefore their opinions about themselves and their behaviour can be biased.
Since experts are not consumers themselves, their opinions about the target market are unbiased. Fertile grounds for experts are universities, institutions and media. Lead users of a product are also consulted as they develop a sophisticated understanding of the product and the uses it can be put to.
They are also able to predict the trajectory that consumer needs would take, and the type of technologies that would be needed to serve those needs. They are way ahead into the future of the markets and technologies that they are experts of, and companies would do well to tap their insights.
Observation is an unintrusive method of studying consumer behaviour and is extremely useful when the agency is unfamiliar with the research area. Observation studies are usually conducted in retail stores to track consumer purchases. Various data such as the amount of time spent by the customer, number of choices considered, amount of the product purchased, shopping influences, amount of money spent, reasons for selection or rejection, etc., can be observed. Much depends on the skills of the researcher in gathering and interpreting data.
Descriptive research describes something. Such research may be, for instance, meant to describe customers’ beliefs, attitudes, recall of advertisements and knowledge about its content. However such researches are not merely fact finding exercises, but are meant to gather data for specific objectives.
Descriptive research differs from exploratory research in terms of the formality and structure with which the research is planned. Data obtained from such researches are usually used in taking marketing decisions. Relationships among various variables are examined in such researches, though cause-effect relationship cannot be established. Descriptive research is quantitative in nature, involving surveys using questionnaires and statistical analysis. Large and representative sample sizes permit generalization of results, aiding in the decision making process, and the collection as well as interpretation of data is objective.
Experimental research establishes cause and effect. It involves setting up of control procedures to isolate the impact of a factor, like price discount on a dependent variable like sales. The idea is to eliminate other explanations of change in the dependent variable. For example, a product is sold on discount in a few stores, and it is sold on undiscounted price in a few similar stores. It is important that customers visiting the two set of stores are similar in as many ways as possible. Statistical significance testing is done to test whether the difference in sales is caused by the discount, or it is a simple random variation.
The main issue in the data collection stage is the technique to collect data from the chosen sample representative.
The sampling process aims at deciding who and how many people should be interviewed. First, the universe, i.e., the group which forms the subject of study has to be clearly defined. The universe depends on the objectives of the study. The sampling frame must be chosen next. This comprises the list or other records of the chosen universe from which a sample can be selected.
Next, it is important to determine the sample size. This involves arriving at the number of respondents who must be surveyed to yield a representative sample of all demographic subgroups of respondents who are being studied. Though it is expensive to interview a large sample size, the larger the sample size, the more is the sample representativeness of the population, and more strongly can the results of the survey be extrapolated to the population. Best results are obtained by interviewing everyone in the population.
The agency has to be mindful of sampling error, which is the error caused by not interviewing everyone in the population. The number of people interviewed is based on a balance between sampling error and cost—the more the number of people interviewed, the less the sampling error, but more is the cost of conducting the survey and vice versa.
After selecting the sample size, it must be determined as to how the sample will be selected for response. The sample can be selected by using either the probability methods or by using the non-probability methods. The probability method is used when every sampling unit has an equal probability of being selected. The selection of every item in the sample is independent of the person who conducts the survey. There are three probability sampling methods:
i. Simple random sampling:
The agency assigns a number to each individual in the sampling frame, and the numbers are drawn at random to complete the sample. Therefore, everyone on the list has an equal probability of being part of the sample.
ii. Stratified random sampling:
The agency divides the population into groups, such as those on the basis of age, gender or income and respondents are chosen randomly from each group. The number of respondents from each group corresponds to the size of the group, i.e., if the population has twice the number of youths than old people, the sample would also have twice the number of youths than old people. The method ensures that each group gets representation in the sample corresponding to its size in the population.
iii. Cluster sampling:
The population is divided into mutually exclusive groups, such as residential blocks, and the researcher randomly selects residential blocks to be interviewed.
In the nonprobability sampling methods, every sampling unit does not have an equal probability of being selected. Thus, there is some researcher bias in selecting the sample item. There are three nonprobability sampling methods:
i. Convenience sample:
The researcher selects the most easily available sampling units or respondents from the population and interviews them.
ii. Judgement sample:
The researcher uses his judgement to select population members from whom appropriate information can be obtained.
iii. Quota sampling:
There is no sampling frame, but the agency knows the percentage of population that lies in various groups in terms of age, gender or income. The sample is made by selecting individuals on the basis of this percentage. For instance, a sample will contain roughly 50:50 females to males, because that is roughly the ratio of males and females in the population.
This is a non-random method, because all members of the population do not have an equal chance of selection—the interviewer selects members according to his convenience and availability of members, but ensures that the ratios among groups are maintained.
This method is adopted when the population is widely dispersed. The difference between quota sampling and stratified sampling method is in the method of selection of the sampling units. In the former, the sampling units are selected by non-random methods such as by convenience or judgement, while in the latter, they are selected on a random basis.
Survey method involves determining how to interview the respondents who have been selected.
Response rates are higher as the personal element in a face to face interview makes refusal to respond less likely. Such interviews can be self-administered, i.e., completed by the respondent himself, or administered by the researcher.
In sample surveys, personal interviews are used frequently. Probing of the respondent is easier with face to face interviews. Clarifying probes help interviewers to understand what the interviewee is saying. Exploratory probes stimulate interviewees to give full answers.
Probing is possible but restricted due to time pressures and a less personalized situation in telephonic interviews. Visual aids can also be used in face to face interviews. However, interviewer bias may creep in during selection of respondents and administration of the questionnaire. Preparing a structured questionnaire can eliminate some element of the interviewer bias.
Experimental design, i.e., testing effectiveness of a stimulus would normally be conducted by this method, rather than by a mail survey, where there is high non-response rate and lack of control over who completes the questionnaire, which could invalidate the results. Use of many open-ended questions in a mail survey would lower response rates, and time restrictions for telephone interviews limit their use for conducting interviews.
But personal interviews are expensive. The presence of an interviewer can cause bias, i.e., the respondent may give socially desirable answers and that may lead to misreporting of sensitive information.
Telephonic interviews have response rates and cost that lie in between those of face to face interviews and mail surveys. Telephonic interviews allow some degree of flexibility when interviewing, as there is a two way flow of information between the interviewer and the respondent, which is absent in mail surveys, but this flexibility is lesser than in personal interviews. Use of visual aids is not possible. There are limits to the number of questions asked before the respondent either terminates the interview or gives quick and sometimes invalid answers to speed up the process.
The use of computer aided interviewing is growing. The presence of accurate databases helps telephonic interviews immensely. Certain type of surveys, for instance, feedback studies, and studies that explore prospective customers, can be done by using telephones.
These are least expensive and can cover widely dispersed population. Response rate is very low and there is danger of unrepresentative sample. The questionnaire is fully structured and probing is not possible. Control over who fills the questionnaire is low. Visual aids can be supplied and because of self completion, the interviewer bias is less.
Response rate can be increased by prior notification by telephone, monetary and non-monetary incentives, by providing stamped return envelopes, granting anonymity to respondents, close-ended questions in the questionnaire and follow up telephone calls.
Three conditions are necessary to get a true response to question-respondents must understand the question; they must be able to provide information and must be willing to provide it. Questions need to be framed in the language that the respondent understands. Researcher must not ask about issues that respondents cannot remember, or are outside their sphere of experience. Researchers need to consider the best way to elicit sensitive or personal information.
The questionnaire may be self-administered, i.e., the respondent fills up the questionnaire himself, or it may be in the form of an interview whereby the researcher asks questions of the respondent. Whatever be the mode of administration, the sequence of questions, wording, number of questions and the length of the questionnaire, type of questions (open or close-ended questions or statements), as well as the techniques used to measure the responses affect the quality of the information obtained.
i. Ordering of topics:
The questionnaire starts with questions which the respondents might find easy to answer. Respondents do not want to sound ignorant, and if they are confronted with a question that they have a difficulty in answering right in the beginning of the interview, they may just decline to go further. The interviewer asks awkward questions before attitude measurement. He asks questions related to unaided awareness before aided ones. Classified information such as age, occupation should always be asked at the end.
ii. Type of questions:
Close-ended questions limit the range of answers that respondents can give. They have to choose their answers from the options given in the questionnaire, which may not be their true answer. Open-ended questions allow respondents to answer questions in their own way. They can elaborate and explain.
iii. Take care in wording of questions:
Guard against ambiguity, leading questions, asking two questions in one and using unfamiliar words. Instructions should be printed in capital letters or underlined so that they are easily distinguished from questions.
The questionnaire should not appear cluttered. In a mail questionnaire, it is a mistake to squeeze in too many questions in one page to reduce the number of pages. Response is likely to be lower if the questionnaire is heavy, than if the number of pages is extended.
Attitudes and beliefs can be measured by means of scales, for instance, a scale ranging from ‘Strongly agree’ to ‘Strongly disagree’.
vi. Probes and prompts:
They seek to explore or clarify what a respondent has said. Respondents may use vague words or phrases. ‘Probe’ is used to get clarifications, and ‘prompts’ are used to elicit responses to a question.
Once the preliminary questionnaire has been designed, it should be tested with a representative subsample to identify faults. Piloting tests questionnaire design and helps to estimate costs. If the pilot proves satisfactory, the final questionnaire is administered to a chosen sample. Usually, the reliability and validity of the instrument is checked during the pilot stage. Other problems such as response and non-response errors, the ability of respondents to understand the instrument, length and time taken to complete the instrument are assessed in this stage.
If the first pilot study identifies some drawbacks in the instrument, i.e., the questionnaire, a second pilot may have to be conducted to rectify the errors and retest the new questionnaire that has been formulated.
Before analysis and interpretation, the data has to be prepared. The raw data that has been collected must be edited. Preliminary checks must be conducted to improve the quality of the data collected. It must also be verified that the sampling instructions have been adhered to. Such checks must be done during the data collection process as well as at the end. Thereafter, the data has to be codified and tabulated. These procedures must be done by those who are familiar with the entire research, from the objectives to the research design.
Basic marketing research analysis can be carried out using software analysis packages. Sophisticated analysis is conducted using packages such as SPSS. Analysis may be carried out to find mean, standard deviation, frequency tables, cross tabs and to conduct other advanced statistical analysis.
Several sophisticated statistical analysis may be used to formulate decisions such as segmentation of the relevant target audience, identifying key factors in new product launches, tracking brand image, etc.
An agency needs to exercise caution when it is interpreting the results of its marketing research endeavour. Agencies commonly infer cause and effect, when only association has been established by the data.
For example, data may show that sales rise when number of salespeople is increased, but it does not mean that increasing the size of the sales force will necessarily lead to increase in sales. The company’s increased sales might have been the result of increased advertising spend, higher margin for retailers, or even lackluster performance of a competing brand.
An agency also needs to exercise caution when it is interpreting means and percentages. Since in most marketing research projects, a sample is surveyed and not the whole population, a mean or percentage is just an estimate, subject to sampling error, i.e., the error due to taking a sample rather than interviewing the entire population.
Statistical hypothesis testing allows sampling differences to be evaluated in the light of sampling errors to establish whether they are likely to be real differences, i.e., statistically significant differences, or they are likely to be a result of taking a sample rather than interviewing the entire population.
The results of the research have to be presented in the form of a report or a presentation.
The key elements in a research project are:
i. Title page, containing the topic of the research study, time of the study, to whom is the report being submitted to and by whom has the research been done.
ii. List of contents, containing a detailed list of the contents of the report along with page numbers, list of tables and list of figures in die report.
iii. Preface, comprising outline of clients’ brief, statement of objectives, scope and methods of research.
iv. Executive summary, comprising a summary of conclusions and recommendations. The purpose of the executive summary is to summarize the key issues and findings of the report for executives who do not have the time to read the complete report.
v. Previous related research, detailing how previous research has had an influence on this research. This section should list out the methodology, context, results and conclusions of related researches done earlier to provide a relevant perspective for the present study.
vi. Research methods, which explains the methodology used for the purpose of the present research. This includes details about the type of study, sampling issues, data sources, data collection methods, use of instruments, preparation of the instruments and method of administration.
vii. Research findings, which should detail the results of the study according to the objectives with which the study was initiated.
viii. Conclusions of the study, along with suitable recommendations.
ix. Appendices, including reports, articles, tables and figures that are not directly related to the study, or can interrupt the flow of the report, though they are useful as background information.
The report should be written in the language which the reader will understand. Jargon should be avoided.
The most important parts of the report are the findings and conclusions based on the survey. These should be presented in a clear and lucid manner. As far as possible, the results should clearly point to some outcomes or suggestions that can help the management in taking decisions or solving the problems that were outlined in the objectives of the study. Sometimes results may point towards additional research being required.