1. Generic vocabulary & useful concepts to know
1.1. Marketing as a discipline
1.1.1. Strategic marketing
1.1.1.1. Segmenting
1.1.1.2. Positioning
1.1.1.3. Targeting
1.1.2. Research Marketing
1.1.2.1. Qualitative
1.1.2.2. Quantitative
1.1.3. Operational marketing
1.1.3.1. Product
1.1.3.2. Place
1.1.3.3. Price
1.1.3.4. Promotion
1.2. Churn analysis: to churn, is when a customer leaves a company. We conduct chrun analysis to identify the possible predictors to identify in advance the consumers who are most likely to churn.
1.3. Direct vs indirect competitor
1.3.1. An indirect competitor satisfies the same need with a different product or solution
1.3.1.1. Coca Cola and water both satisfy thirst
1.4. SBU: means a strategic business unit (a product division, or activity)
1.5. Harvard Blue ocean versus Red ocean strategy
1.5.1. Blue oceans denote all the industries not in existence today – the unknown market space, unexplored and untainted by competition
1.5.1.1. These are innovative and new markets
1.5.2. Red oceans are all the industries in existence today – the known market space, where industry boundaries are defined and companies try to outperform their rivals to grab a greater share of the existing market.
1.5.2.1. Crowded markets
1.6. Indicators and metrics related to
1.6.1. opinions
1.6.1.1. SOFT METRICS
1.6.2. Behavior and transactions
1.6.2.1. HARD METRICS
1.7. A concept is an intellectual abstraction. We use concepts to understand the world around us and build theories and relationships that explain our environment.
2. 4- Data analysis
2.1. Descriptive statistics
2.1.1. Definition: Descriptive statistics make it possible to enumerate and describe a population in an exhaustive way (eg success rate).
2.1.1.1. We distinguish between
2.1.1.1.1. Univariate statistics
2.1.1.1.2. Bivariate statistics
2.1.1.1.3. Multivariate statistics
2.1.2. A-Univariate statistics
2.1.2.1. Categorical variable
2.1.2.1.1. You can create frequency tables for that type of variables (see slide 34)
2.1.2.1.2. Graphical representation: we use bar charts or Pie charts. See slide 43
2.1.2.2. Scale variable
2.1.2.2.1. You can compute
2.1.2.2.2. Graphical representation:
2.1.3. B-Bivariate statistics
2.1.3.1. Categorical variable
2.1.3.1.1. We can create a contingency table (2 categorical variables) -- see slide 58
2.1.3.1.2. Graphical representation: you can use bar plots
2.1.3.2. Crossing a scale variable and categorical variable (aka grouping variable) - slide p65
2.1.3.3. Crossing two scale variables
2.1.3.3.1. Variables
2.1.3.3.2. We can observe
2.1.3.3.3. Graphical representation: we use a scatter plot (correlation plot) to represent the relationship between two variables. See slide 68
2.2. Inferential statistics
2.2.1. Definition: It is about estimating the numerical characteristics of a population from the observation of one of its parts (a sample) thanks to the inference.
2.2.1.1. We need inference to express with a certain degree of confidence that (a difference, or relation) we observed within our sample will probably be observed in the population if we conduct a census.
2.2.2. We use hypothesis testing to infer if what we observed in our sample can be generalized to the population with some level of confidence
2.2.2.1. All the inferential tests work with the same logic
2.2.2.1.1. A hypothesis is a quantitative statement about the characteristics of a population.
2.2.2.1.2. We distinguish H0 or null hypothesis and its alternative hypothesis called H1
2.2.2.1.3. To decide whether or not to reject H0 (hence its "plausible" character), we examine the statistical significance of the test: p-value. --- See slides 80-83
2.2.2.2. What types of tests exist ? (toolbox)
2.2.2.2.1. We distinguish between parametric vs non parametric tests
2.2.2.2.2. How to select the right test ? (see slide 85 and 90) for details
3. 3- Quantitative methods
3.1. Questionnaire design
3.1.1. Generic rules
3.1.1.1. Funnel rule
3.1.1.1.1. going from general to specific questions
3.1.1.2. Don t design complex, long questionnaires
3.1.1.3. The filter questions should be in the beginning
3.1.1.4. The identification questions at the end of the questionnaire
3.1.1.5. A modality: means a possible answer
3.1.1.5.1. For example if the possible answers are YES or NO for a question, then we say that the question (or variable) has 2 modalities (meaning two possible answers)
3.1.2. Content
3.1.2.1. List the objectives and sub objectives or your research then transform them into questions
3.1.2.1.1. Vocabulary
3.1.2.1.2. Style of redaction
3.1.3. Type of questions
3.1.3.1. VERY IMPORTANT BECAUSE IT DETERMINES THE TYPE OF POSSIBLE STATISTICAL ANALYSIS TO CONDUCT
3.1.3.2. Type of questions
3.1.3.2.1. Open questions
3.1.3.2.2. Closed questions
3.1.3.2.3. Specific categories questions
3.1.3.3. Measurement level (what modalities express)
3.1.3.3.1. Categorical
3.1.3.3.2. Quantitative or scale
3.1.3.4. Coding the questionnaire
3.1.3.4.1. answers are represented as numbers to be analysed by the software
3.1.3.4.2. Text is coded via content analysis then it s analyzed via software
3.1.4. Data collection methods
3.1.4.1. CAPI
3.1.4.1.1. Computer Assisted Personal interview
3.1.4.2. CATI
3.1.4.2.1. Computer Assisted Telephone interview
3.1.4.3. CAWI
3.1.4.3.1. Computer Assisted Web interview
3.1.5. Rules for a good questionnaire (addendum)
3.1.5.1. Questions
3.1.5.1.1. No double meaning questions
3.1.5.1.2. No ambiguous words
3.1.5.1.3. Identification questions (for example: gender, age, etc) should be at the end of your questionnaire. The only exception is if they are used as filter questions (like a quota, or to select the respondents you want to survey)
3.1.5.2. Modalities
3.1.5.2.1. Must be exclusive
3.1.5.2.2. A homogeneous number of modalities for Likert scale (5 or 7, etc.. )
3.2. Selecting the sample
3.2.1. Vocabulary
3.2.1.1. Reference population
3.2.1.1.1. “The set of objects having the desired information to meet the objectives of a study”.
3.2.1.2. Sampling frame
3.2.1.2.1. is a concrete representation of the elements of the target population (For example the list that contains all the reference population)
3.2.1.3. A sample
3.2.1.3.1. "is a subset of elements extracted from a population they are supposed to represent"
3.2.1.4. Sampling unit
3.2.1.4.1. It is an "object, or entity containing an object, that will be selected for inclusion in the sample"(individual? Household? state? etc.)
3.2.1.5. Unit of analysis
3.2.1.5.1. It is an "object, or entity containing an object, that will be selected for the analysis“
3.2.1.6. We distinguish between
3.2.1.6.1. The census
3.2.1.6.2. The sample
3.2.2. What errors to avoid when sampling? There are 2 types of possible errors
3.2.2.1. Sampling error
3.2.2.1.1. the fact of forgetting one or more important observations. which make the sample unrepresentative, such as not taking Renault-Nissan into account in a study of the French automotive industry.
3.2.2.2. The non sampling error
3.2.2.2.1. it is, by contrast, any error that is not due to sampling (For example: a response error, input, etc.). It's an error that results during data collection, causing the data to differ from the true values.
3.2.3. There are basically 2 sampling approaches (to select a sample)
3.2.3.1. Probabilistic methods (we randomly select respondents). Each member of the population has a known non-zero probability of being selected. YOU SHOULD GENERALLY HAVE A SAMPLING FRAME (LIST THAT CONTAINS ALL THE POPULATION)
3.2.3.1.1. 1- Simple random survey (sampling)
3.2.3.1.2. 2- Random systematic survey (sampling)
3.2.3.1.3. 3- Stratified sampling: A stratum is a subset of the population that share at least one common characteristic. Examples of stratums might be males and females, or managers and non-managers. The researcher first identifies the relevant stratums and their actual representation in the population. Random sampling is then used to select a sufficient number of subjects from each stratum. "Sufficient" refers to a sample size large enough for us to be reasonably confident that the stratum represents the population. It can be Proportional or Non proportional
3.2.3.1.4. 4- Cluster survey (sampling)
3.2.3.2. Non probabilistic methods (A.k.a. empirical methods). Members are selected from the population in some nonrandom manner (quotas, convenience, etc )
3.2.3.2.1. Convenience Sampling
3.2.3.2.2. Judgment (purposive) sampling
3.2.3.2.3. Referral /Snowball Sampling
3.2.3.2.4. Quota Sampling
3.3. Sample size
3.3.1. To compute the sample size we need three elements:
3.3.1.1. E: desired margin of error (precision). If you want to augment your precision, you will have the augment the sample size and thus the cost.
3.3.1.1.1. Having a margin of +/-3% means that if you have a result of 60% in your sample, it means that the real result in the population is probably (see Z below) between 57% and 63%
3.3.1.2. Zc is the critical value from the table of probabilities of the standard normal distribution for the desired confidence level (generally Z=1.96 for 95% confidence)
3.3.1.2.1. You will express your results with a confidence of 95% that your estimations for the reference population are correct based on what you observed in your sample (60%), so the real score is somewhere between {57%-63%} and we are 95% confident about that result.
3.3.1.3. p (for a percentage) or σ (for an average). p is the proportion of a phenomenon in the case of an estimated percent (for example, we want to estimate the % of people that vote republicans). Meanwhile, σ is the standard deviation in the case of estimating an average (for example, we want to estimate the monthly leisure budget of our respondents)
3.3.1.3.1. If you don't have 'p' you have 2 choices
3.3.2. Some statistical concepts
3.3.2.1. The standard deviation
3.3.2.1.1. The Standard Deviation is a measure of how spread out numbers are. We can say about it that it's the average deviation from the average (or mean). Standard deviation is calculated as the square root of variance
3.3.2.2. Variance
3.3.2.2.1. The average of the squared differences from the Mean. You can look on blackboard for the file: Computing variance and standard deviation.xlsx
3.3.2.3. The Z score
3.3.2.3.1. Z score indicates how many standard deviations an element is from the average. The formula is (x-average/(standard deviation)) -- see slides - P34
3.3.2.4. The normal distribution
3.3.2.4.1. The normal distribution (Or Gaussian) is a very common distribution for quantitative values. The normal distribution is the most important distribution in statistics because it fits many natural and social phenomena. For example, errors in measurements, blood pressure, marks on a test, etc. The normal distribution describes how the values of a variable are distributed.
3.3.2.4.2. Addendum: there are many other statistical distributions (t distribution, binomial, etc), but generally, when we have a sample size bigger than 30 observations, we start to observe that they tend to converge towards a normal distribution. This is called the central limit theorem.
4. 2- Qualitative research approach
4.1. History
4.1.1. Methods emerged from psychological tools used in the last century
4.1.2. CCT a stream of research that studies the symbolic status of consumption
4.1.3. Experiential stream of research studies the emotional and sensory aspects of consumption (humans are irrational)
4.2. Methods used to collect data with the qualitative approach
4.2.1. Individual interview
4.2.1.1. can be
4.2.1.1.1. Non directive (clinical)
4.2.1.1.2. Semi directive (using an interview guide)
4.2.1.1.3. Directive (a questionnaire with open questions)
4.2.1.2. Sample size: we use the saturation criteria
4.2.1.3. Verbatim: the full transcription of the interview -- the text and material that will be analyzed --
4.2.2. Group interview (Focus Group)
4.2.2.1. Great for finding solutions to a business challenge + fast + not expensive
4.2.2.2. Be cautious about the tendency of the respondants to conform (asch conformity)
4.2.3. Projective techniques: are indirect methods used in qualitative research. These techniques allow researchers to tap into consumers' deep motivations, beliefs, attitudes and values
4.2.3.1. using metaphors, sometimes pictures or any kind of material to understand the associations, motivations and beliefs of a customer
4.2.3.1.1. The chinese protrait game: If this brand was an animal, what type of animal it would be?
4.2.3.1.2. Select the pictures that reflect the best this brand?
4.2.4. Observation methods
4.2.4.1. Direct
4.2.4.1.1. Ethnography
4.2.4.1.2. Garbology
4.2.4.1.3. Netnography
4.2.4.2. Indirect
4.2.4.2.1. Netnography
4.2.4.2.2. Physiological measurment
4.3. Methods used to analyze qualitative data
4.3.1. Content analysis (thematic analysis)
4.3.1.1. Manual
4.3.1.1.1. Extracts categories and sub categories of meaning expressed by customers and places them on a grid (table) to make them easier to read and count their occurences
4.3.2. Text mining (lexicometric analysis)
4.3.2.1. Automated (supervised)
4.3.2.1.1. The software analyses the corpus (or verbatim)
5. 1- Marketing research in general
5.1. Consumption value framework
5.1.1. Utilitarian: objective consumption
5.1.2. Experiential: about sensory and emotional elements
5.1.3. Symbolic or conspicuous
5.1.4. Aesthetic or consuming for moral values (CSR, spirituality, etc)
5.2. Marketing problem (symptoms) Vs. Research problem (underlying causes)
5.3. The difference between a conjecture and a hypothesis
5.3.1. Conjecture (Full list of possible explanations)- it s a non mature hypothesis
5.3.2. Hypothesis: short list of possible explanations to test
5.4. Marketing research process 6 stages
5.5. Marketing research can be used both for
5.5.1. Problem indentification
5.5.2. Problem solving research
5.6. 2 Sub-branches of marketing research
5.6.1. Examples of common quantitative methods used (conclusive)
5.6.1.1. Segmentation
5.6.1.1.1. Also known as Clustering
5.6.1.1.2. 5 segmentation criteria
5.6.1.1.3. Famous algorithms
5.6.1.2. Positioning maps
5.6.1.2.1. Used to understand how a consumer perceives a brand or product compared to others (Volvo is associated with safety for example)
5.6.1.3. Conjoint analysis
5.6.1.3.1. Breaks down a product into attributes and levels
5.6.1.3.2. The method allows computing indirectly preference parthworths (importance that a consumer gives to a specific level or attribute)
5.6.1.3.3. Limitation: the method is not useful for sensory products like perfume, etc..
5.6.1.4. Choice modeling
5.6.1.4.1. Predicts a discrete choice of a customer
5.6.1.5. Pricing (Gabor Granger method)
5.6.1.5.1. A method that asks a customer the likelihood of his purchase based on various price levels proposed to him.
5.6.1.6. Targeting
5.6.1.6.1. BCG Boston Consulting Group
5.6.1.6.2. GE McKinsey matrix
5.6.1.7. Resource allocation
5.6.1.7.1. A ressource can be money, time or effort
5.6.1.7.2. Effectiveness Vs. Efficiency
5.6.1.7.3. The method answers 2 questions
5.6.1.8. Customer lifetime value (a method used in customer relationship management)
5.6.1.9. Bass forecasting method
5.6.1.9.1. Forecast sales timeline and adoption rate for innovations
5.6.2. Examples of qualitative approaches (exploratory)
5.7. Types of data collected by marketing research
5.7.1. Informations about
5.7.1.1. Demand: describes the buyers in general, customers opinions, behaviors, etc
5.7.1.2. Offer: describes the competitors, the available products on a market, etc
5.7.1.3. Environment
5.7.1.3.1. Using PESTEL framework before the SWOT analysis
5.7.2. Internal vs External data
5.7.2.1. Internal
5.7.2.1.1. CRM customer relationship management system
5.7.2.1.2. Marketing information system, etc
5.7.2.2. External sources
5.7.2.2.1. Associations, international organisations, etc
5.7.3. Secondary Vs. Primary data
5.7.3.1. Check-list before conducting marketing research
5.7.3.1.1. Before conducting qualitative or quantitative research we conduct Desk Research (less expensive) and helps identify if the data exists somewhere
5.7.3.1.2. Is the issue that important to invest in marketing research?
5.7.3.1.3. Hidden motives: political reasons to justify a decision
5.7.4. Periodicity
5.7.4.1. Ad Hoc or cross-sectional
5.7.4.2. Follow-up
5.7.4.3. Longitudinal - panel data -
5.8. Research design; is the plan or approach selected to answer your research problem
5.8.1. Type of research design
5.8.1.1. Desk research
5.8.1.1.1. Using available secondary data and databases: it s faster and less expensive
5.8.1.2. Qualitative
5.8.1.2.1. Individual interview
5.8.1.2.2. Group interview (focus group)
5.8.1.3. Quantitative
5.8.1.3.1. Cross-sectional
5.8.1.3.2. Longitudinal - Panel -
5.8.1.3.3. Experimental
5.8.1.4. Mixed Design (most of the time we have a qualitative, then a quantitative phase)
5.8.1.4.1. Triangulation approach
5.8.2. Research design objectives
5.8.2.1. Exploratory research
5.8.2.1.1. Mostly qualitative or Desk research
5.8.2.1.2. Subjective
5.8.2.1.3. Not generalizable
5.8.2.2. Descriptive research
5.8.2.2.1. very broad applications
5.8.2.2.2. Describe behaviors, opinions, etc
5.8.2.3. Causal (experimental research)
5.8.2.3.1. The most ambitious and complex type of research objective
5.8.2.3.2. Verifies the existence of a cause and effect relationship between a predictor variable (independent) and an outcome variable (dependent)
5.9. Marketing research documents
5.9.1. Research brief
5.9.1.1. The manager explains his managerial problem, the context and suggests an approach and conditions of execution (time and budget)
5.9.2. Research proposal
5.9.2.1. The marketing researcher answers the research brief by suggesting the most appropriate research design, and specifies the timeline of execution.
5.9.3. The instrument used to collect data
5.9.3.1. Qualitative approach
5.9.3.1.1. interview guide
5.9.3.2. Quantitative approach
5.9.3.2.1. Questionnaire
5.9.4. Analysis plan
5.9.4.1. a document that describes the statistical analysis that the research should conduct
5.9.5. Final report