Marketing Research vs Big Data Analytics

Recap & Context

  • Previous video dealt with MIS (Marketing Information System) & the growing role of data in extracting consumer insight.
  • Current focus: contrast Big Data Analytics with Marketing Research—two complementary but distinct information-gathering systems inside most firms.

Big Data Analytics (BDA)

  • Works almost exclusively with secondary, historical, behavioural data found in existing internal or external databases.
    • Typical sources: click-streams, POS logs, loyalty cards, CRM files, accounting ledgers.
  • Goal: prediction – use statistical/AI tools to forecast future behaviour from past patterns.
    • Conceptual model: \text{Past Behaviour} \xrightarrow[\text{stats\/ML}]{\text{model}} \text{Probability of Future Event}
  • Position inside firms
    • Often its own analytics division separate from Marketing Research.
    • Overlap exists at the analysis phase (both sides may run regressions, clustering, ML, etc.).

Marketing Research (MR): Definition & Purpose

  • Formal definition: a systematic design, collection, analysis & reporting of data relevant to a specific marketing situation facing the organisation.
  • Key distinction from BDA
    • MR emphasises primary data – information collected now for a particular decision.
    • Can still mine secondary data, but only as one tool among many.
  • Example: measuring current customer satisfaction cannot rely only on click-stream history → must directly ask customers.
  • Departmental realities
    • Large firms: separate MR and BDA departments within marketing.
    • Small firms: may have no in-house MR; outsource to research agencies or buy syndicated/industry reports.

The Marketing Research Process (5-Step Road Map)

  1. Define the Problem & Research Objectives
    • Clarify what question must be answered (e.g., “Why is churn up 5 %?”).
  2. Develop the Research Plan – the “travel itinerary” analogy (trip from SF → LA).
    • Select research approach (exploratory, descriptive, causal).
    • Draft sampling plan (who, how many, selection method).
    • Choose contact methods & research instruments; compile into a research proposal (includes timeline & cost).
  3. Implement the Plan & Collect Data
    • Fieldwork: surveys, interviews, observations, experiments, etc.
  4. Analyze & Interpret Data
    • Statistical tests, modelling, visualisation – shared tool kit with BDA.
  5. Report Findings & Recommendations
    • Translate numbers into managerial action items.

Research Approaches Explained

  • Exploratory Research – open-ended “sense-making” to refine the problem.
    • Techniques: literature search, secondary-data scan, depth interviews.
  • Descriptive Research – answers who/what/where/when/how; still not why.
    • Example: Target’s motherhood-prediction used shopping baskets to describe who buys unscented lotion, when, in what quantities.
  • Causal Research – establishes cause → effect (the why).
    • Requires controlled experiments.
    • Email-timing example: send identical content Mon/Wed/Sat, hold everything else constant; measure lift.

Data Types

  • Secondary Data (already exists)
    • Pros: cheap, quick, sometimes impossible to collect otherwise.
    • Cons: risk of irrelevance, inaccuracy, outdated definitions, wrong aggregation level.
  • Primary Data (collected now for this study)
    • Requires planning across: approach, contact method, sampling, instrument design.

Contact / Collection Methods

  • Observation – watch behaviour without questioning.
    • Includes ethnography (researcher embedded in consumer’s natural setting).
  • Survey – ask respondents about knowledge, attitudes, preferences.
    • Modes: mail (rare today), telephone, personal interview, online (dominant).
  • Experiment – manipulate independent variable(s) → observe effect on dependent variable(s).

Cost & speed hierarchy (cheapest/fastest to dearest/slowest):
\text{Online} < \text{Mail} < \text{Telephone} < \text{Personal Interview}

Sampling Fundamentals

  • Purpose: study subset (sample) to infer about whole population.
  • Key design questions
    1. Who is to be surveyed? (target population)
    2. How many? (sample size nn)
    3. How selected? (probability vs non-probability technique)
Probability Sampling (each unit has known p=1Np = \frac{1}{N})
  • Simple Random – pure lottery.
  • Stratified – sample within predefined strata (e.g., class year).
  • Cluster – sample entire clusters (e.g., ZIP codes) then possibly every member within cluster.
Non-Probability Sampling
  • Convenience – intercept survey outside student union.
  • Judgment – expert picks “representative” cases.
  • Quota – force proportions (e.g., 30%30\% seniors, 25%25\% sophomores, 45%45\% freshmen).

Research Instruments

  • Questionnaire – core MR instrument.
    • Closed-ended items (MCQ, Likert scales) vs Open-ended (verbatim comments).
  • Mechanical Devices – traffic counters, eye-tracking cameras, in-store RFID, web cookies.

Rise of Online & Social Tracking

  • Web analytics capture full click-paths from ad impression → conversion.
  • Online Listening – monitor consumer chatter on Facebook, Instagram, Twitter.
  • Social Targeting – leverage individuals’ network ties & online conversations for personalised offers.
  • Virtual focus groups, bulletin boards, A\/B testing tools make research faster & cheaper.

Data Analysis & Intersection with Analytics

  • MR analysts increasingly deploy big-data tool kits (Python/R, machine learning) once primary data are gathered.
  • Thus, MR + BDA converge at analytics stage but diverge in data origin & purpose.

Ethical & Practical Considerations

  • Privacy – tracking & social targeting raise consent and data-security issues.
  • Data Quality – secondary data may be outdated; primary surveys risk measurement error.
  • Representativeness – non-probability samples threaten external validity.
  • Resource Limits – small firms often outsource MR to control cost.

Comparative Summary: MR vs BDA

  • Data Source
    • MR: Mainly primary (can include secondary).
    • BDA: Secondary/historical by default.
  • Primary Goal
    • MR: Solve specific problem today.
    • BDA: Forecast future outcomes.
  • Method
    • MR: Follows 5-step research process; gathers new data via surveys, experiments, etc.
    • BDA: Mines existing large datasets; emphasises algorithmic pattern recognition.
  • Organisational Setup
    • Often separate teams; collaboration at analysis phase common.

Key Takeaways

  • Marketing Research provides a flexible, problem-focused framework for collecting new information when existing data are insufficient.
  • Big Data Analytics excels at large-scale pattern detection & prediction using existing behavioural traces.
  • Together they enable evidence-based marketing decisions: MR clarifies what to ask, BDA helps scale insights & anticipate future behaviour.