Lecture 3: The Consumer Decision Process

The Consumer Decision Process: Outcomes and Products

• Consumers don’t buy products → They buy the outcomes they give

  • (Shoes → Status, Comfort, Performance)

A quick summary (You already know all of this) : 1) Customer Value Formula 2) Benefits 3) Costs

The key point about features vs benefits

• CV = Perceived Benefits - Perceived Costs (How Consumers make decisions)

• 4 Types of Benefits/Value: 1) Social 2) Functional 3) Economic 4) Experiential

• Cost Types: Time, Risk, Switching, Learning (Incl. Non Monetary)

• Key Point: Features are Comparable → Benefits are Differentiation

Understanding Value, How we can observe it

• Value shows itself in tradeoffs and decisions: Purchasing, Switching Brands etc.

• Observable in demand curves: High Value = Less Price Sensitivity (Inelastic Demand)

  • TLDR: Raise Prices → No move in demand = Value Advantage

Defining: Need, Preference and Value

• Needs (Arena): The basic problem to solve —>Transport from NYC to Washington

• Preference (Segment): Desired way to solve problem —> Train or Car

• Value (Segment Winner): Which has higher perceived net benefit → Acela or Regional train

4 Types of Customer Needs

• Explicit

• Latent

• Stated

• Unstated

Explicit Needs

• Customer Clearly expresses (Identifiable through survey)

  • I want noise cancelling headphones

Latent Needs

• Customer isn’t aware of them, cannot articulate (Discovered through observing behavior)

  • Busy Users appreciate face-id

Stated Needs

• What Customer says they want (Direct Input)

  • I want a vegan fast food option

Unstated Needs

• Customer expects this, doesn’t state it but it’s vital for their satisfaction

  • Clean Restaurant

Differentiation

• Innovation/Differentiation through fulfilling Unstated and Latent Needs → Leads to delight

  • Best companies know what consumer wants → Before they know it themselves (Apple)

AMA Definition for the Marketing Research Process

• The function that links the 1) consumer, customer and public to 2) the marketer 3) through information

What does Marketing Research allow Organizations to do:

• Get Deep insights into customers → Motivations, Behaviors

• Assess Market Potential (Size and Growth) + Market Share

• Measure how effective our Marketing activities are

• Understand the Marketing Process → Improve how we think about it

The 5 Steps of Marketing Research

1. Define the Problem and Research Objectives

2. Develop a research plan for collecting information

3. Implement the research plan + collect/analyze data

4. Interpret + Report the findings

5. Turn Insights into Marketing Action

Step 1: Defining The problem correctly, Asking the right Research Question + Objectives

• This is the foundation and guides the entire process

  • Clearly Define the problem → Focused Research

  • Questions are specific (focused) but open enough (New Insights Gained)

  • Set SMART Objectives

3 Types of Research Listed in Order

1. Exploratory

2. Descriptive

3. Causal

Step 1: Exploratory Research

• This is our starting point for getting initial insights

• We use qualitative methods to get broad insights (Focus Groups, Interviews)

  • Example: Why are Vegans not happy with our menu

Step 2: Descriptive Research

• Now that we have some insights, we need to Quantify Market Characteristics (Who Buys, How Many, When…)

  • We can use Surveys and Data

  • Example: What % Of adults in Germany consume vegan meals

Step 3: Causal Research

• Now we test Cause - Effect Relationships: Experiments that isolate a single variable → See if changing it leads to a different outcome

  • Example: If we discount prices by 10% do Vegan Products have more sales

Some Best Practices (Manager - researcher) + Deductive and Inductive + Symptoms and Root Problems

1. Managers know the businesses, Researchers the process → Let them Collaborate

2. Do not Confuse Symptoms (Lower Sales) with the Root Problem (Underlying Cause for Declining Sales)

3. Combine Deductive (Data Driven) and Inductive thinking (Observations)

Step 2: How do we design a good Research plan

• We need to know what type of data we need → Who collects it from whom and how (Data Sources and Qualitative vs. Quantitative)

• Focused and Cost efficient

Primary Data Source (2 Types)

• We collect this directly from specific and tailored questions (Surveys, Interviews)

• It’s very reliable but time consuming and costly

Secondary Data + How most Plans work

• Already Exists (Government Database, Internal Records) →Efficient but less relevant/outdated

• Most plans use secondary data first (Good for Initial Exploration) and then Primary

Quantitative Data

• Structured, Numerical Data with Large Samples (Surveys, Experiments)

• Good at answering (How much, How..) + Measuring Behaviors and Preferences

Qualitative Data

• Unstructured Insights from Focus groups and Interviews

• Great at getting Deeper Motivations and Emotions → Great at getting to the Why answers (Why this products)

Surveys (Data Collection Types)

• Quantitatively collect data from large audience → Measure Preferences and Attitudes

Focus Groups

• Qualitatively speak to small groups and figure out why they think like they do (Motivation, Perception)

• Great at testing Ad Campaigns

Observational Research

• Qualitatively observe Behavior in real world → Identify gaps between stated and real behavior

• Example: How Someone navigates a store → Use it to improve layout

Conjoint Analysis

• Qualitatively understand how customers make trade offs between product features (Weigh Them)

  • What Features Matter, Sensitivity to Price Changes

A/B Testing

• Create two versions (Control: A and B: Treatments) → Change 1 variable and measure Performance difference (Advertisements, Thumbnails)

Sentiment Analysis (Social Arbitrage)

• NLP (Natural Language Processing) To review: Comments, Posts, Reviews at scale

• Good at detecting trends, brand reputation and catching user pain points

Implementation Basics + Preview of the 3 main topics

• Resource Intensive -> We are collecting data and processing it for analysis

• Requires Sampling Process, Data Collection Methods/Tools, Ethics and Data Protection

Sample (Sampling Process)

• Subset of total population → Represents whole group

• The goal is Representativeness but it’s tough to figure out who to sample (To mirror target population)

Simple Random Sampling

• Each Individual has equal chance at selection

• Highly Representative but costly → We need accurate list of entire population

Stratified Sampling + Example (Random)

• Population divided into subgroups → Samples drawn from each

• Important subgroups more represented, but complex to design

• Example: Survey equal amount of customers from each age group

Convenience Sampling (Non Random)

• Participants chosen because they are easy to reach → Bias! (Asking around at mall)

Quota Sampling (Non Random)

• Set Quotas for Characteristics (50% Male) → Easy to achieve Representation but prone to slection bias

Choosing the correct data sampling tool: Surveys, Interviews, Observation, Mechanicals

• Surveys (Easy)

• Interviews (Good for Complexity)

• Observation (Gaps between stated-real behavior)

• Mechanical Tools (Scanners, Heat Maps -> Granular Data)

What is Passive Data Collection

• Background Data -> GPS, Website Navigation -> Real Behavior

Data Mining Definition

• Process of Analyzing Large and Complex data sets → Uncover hidden patterns

3 Data Mining Methods

• Market Basket: Customer who buys x, also buys y

• Anomaly Detection: Fraud Detection

• Clustering: Building customer segments based on behavior

Data Quality and how to avoid bad data

• Garbage in Garbage Out → Poor Data = Poor Decisions … Avoid:

  • Pilot Testing → Catch Bad/Biased questions

  • Monitor Poor Results

  • Provide Clear guidance

What is the Goal of Step 4: Interpret and Reporting Findings

• Use data to generate key insights for managers

Big Data + Machine Learning Intro

• Massive Accumulation of Data -> Brings Volume, Velocity and Variety (Large, Fast and Diverse)

• Can Be unstructured → Analyzed via machine learning

Machine Learning 3 Learning Types

• Supervised Learning: Predict outcomes using labeled data (Likelihood to buy)

• Unsupervised Learning: Discover hidden patterns on its own (Hidden Customer Segments)

• Reinforcement Learning: Continuously improves decisions in real time (Better Ad Placement)

Presenting Findings

• Tell a clear and structured story: Problem -> Insight -> Recommended Action

  • Why does this insight matter, what should be done next

4 Pitfalls in Interpretation

• Over Generalizing, Overcomplicating and Misinterpreting data

• Cherry-Picking Favorable Data

Goal of Step 5: Insight to Marketing Action

• Don’t waste potential, move from insight to action

Structuring an Action Plan

• Translate findings into changes in Marketing Mix (4 P’s)

  • Focus on highest Business Impact Decisions

• Set KPI’s (Performance Metrics)

  • Key: Every Insight → Action or Recommendation

4 Misalignments

• Don’t Ignore Insights that contradict internal Assumptions

• Data Decoration: Using selective data to justify pre decided actions

• Skip Testing (No Pilots)

• Misaligned Actions: Not linked to research findings