MARK5827 Customer Analytics
The Complete Study & Assessment Guide · T3 2026MARK5827 Overview
- UNSW Sydney
- Term 3 2026
- Postgraduate marketing
- Four weighted assessments
- Applied R tutorials
MARK5827 Customer Analytics at UNSW Sydney is an applied postgraduate marketing course offered in Term 3 2026. It treats customer analytics as both a scientific and managerial practice.
- Assessment pattern Two Moodle tasks contribute 60%, the group project contributes 30%, and peer evaluation contributes 10%.
- Core habit Define the managerial decision before selecting data, variables or an analytical method.
- Technical pressure Check sample fit, measurement, outliers and scale before trusting a cluster solution.
- Interpretation boundary A segment profile describes a group but does not by itself prove campaign response or causal impact.
How MARK5827 is assessed
| Component | Weight | Format |
|---|---|---|
| Online Quiz | 20% | Moodle task, 20 to 30 minutes, quiz and essay formats, no coding, Weeks 1 to 4 |
| Case Study | 40% | Moodle case study, about 30 minutes, analyst role, Weeks 1 to 6, no coding |
| Group Term Project | 30% | Group report and tutorial presentation, coding required, whole course |
| Peer Evaluation | 10% | About five Moodle questions; submission is mandatory |
The official UNSW Handbook lists Online Quiz 20%, Case Study 40%, Group Term Project 30% and Peer Evaluation 10%. The Term 3 assessment guide supplies the format, dates and topic coverage used here.
Current MARK5827 dates
| Date | Item | Control |
|---|---|---|
| 6 October 2026 | Online Quiz | Scheduled during the lecture; task duration is 20 to 30 minutes. |
| 27 October 2026 | Case Study | Scheduled during the lecture through Moodle. |
| Week 10 | Group presentation | The presentation contributes 40% of the Group Term Project component. |
| 27 November 2026, 5pm | Group report | The report contributes 60% of the Group Term Project component. |
| 30 November 2026, 5pm | Peer Evaluation | Submission is mandatory for this component. |
Dates are as published in the current course materials. Confirm exact deadlines and submission settings in the live LMS.
What MARK5827 covers
The guide follows the detailed Term 3 material available for Weeks 1 to 4: managerial analytics, CRM and customer value, data preparation, RFM and two-step cluster analysis, then latent class analysis and its decision boundaries.
Customer Analytics for Managerial Decisions
Define the decision, prepare relevant data, derive insight, communicate a bounded recommendation and learn from the outcomeCustomer analytics turns customer information into a managerial decision rather than treating data volume or model complexity as the goal.
- Business objectives and measurable success criteria
- Customer data as input rather than automatic insight
- The analytics loop from raw data to action and learning
- CRM, empirical methods and coding as connected capabilities
- Managerial interpretation and communication boundaries
- Quiz, essay and case-study answer structure
CRM, Customer Value and Privacy
Separate customer benefit from provider value, follow the relationship lifecycle and protect trust and data qualityCustomer relationship management selects customers the firm can profitably serve and shapes interactions across the relationship.
- CRM as customer selection and interaction management
- Customer-side functional, economic, emotional and symbolic value
- Provider-side lifetime, referral, influencer and knowledge value
- Customer lifecycle and changing relationship contribution
- Heterogeneity and differentiated CRM interventions
- Privacy, transparency, control and data quality
Data Preparation and Measurement
Test sample fitness, variable meaning, missingness, range logic and categorical recoding before analysisData preparation begins by asking whether the sample and variables can answer the business objective.
- Target population and sample fitness
- Random and stratified sampling choices
- Range, logic and extreme-value checks
- Missing-value deletion and imputation trade-offs
- Nominal, ordinal and metric measurement
- K minus one dummy-variable recoding
RFM and Two-Step Cluster Analysis
Move from RFM rules to hierarchical cluster choice, K-means fitting, validation and actionable profilesSegmentation divides a heterogeneous customer base into groups that are similar within and different across groups. This chapter begins with transparent RFM rules and then develops the course's two-step cluster procedure.
- Segmenting, targeting and positioning roles
- Recency, frequency and monetary-value rules
- Homogeneity, identifiability and sustainability criteria
- Standardisation, outliers and multicollinearity
- Hierarchical elbow and agglomeration changes
- K-means fitting, validation and profile interpretation
Latent Class Analysis and Managerial Action
Select hidden classes with BIC, interpret conditional probabilities and respect the boundary between profiles and targetingLatent class analysis models hidden customer groups through patterns of observed binary responses. Unlike distance clustering, the hidden class is treated as generating different response probabilities on product ownership or behaviour indicators.
- Observed and latent customer heterogeneity
- Legal, economic, technological and innovative constraints
- Binary indicators and hidden class membership
- BIC comparison across candidate models
- Conditional response probabilities and class shares
- Covariates, transitions, response and value boundaries
Students use established empirical methods and R to analyse customer information, but the course consistently returns to the business problem: what decision should a customer relationship manager make, what evidence is relevant, and how can a finding be translated into an action that an executive audience can understand?
The detailed material available for this guide covers Weeks 1 to 4. It begins with the customer analytics process, which links business objectives, raw data, preparation, variable selection, analytical choice, insight and action. It then develops customer relationship management through customer and provider views of value, the customer lifecycle, heterogeneity and privacy.
Data preparation follows, with sample quality, range and logic checks, missing responses, categorical recoding and R workflow. The final teaching block compares RFM segmentation, hierarchical and K-means clustering, and latent class analysis.
Students learn to standardise distance variables, read an agglomeration elbow, return to original units for interpretation, select latent class models with BIC and interpret conditional response probabilities without overstating what the model establishes. The published assessment structure contains two individual Moodle tasks, a group term project and peer evaluation. There is no final exam in the course materials.
Assessment therefore rewards repeated application across the term: choosing a suitable method, interpreting results, evaluating limitations, coding where required and communicating actionable recommendations. This guide stays within the captured Weeks 1 to 4 boundary and uses original examples rather than reproducing official question stems.
Worked example · free
Choose a defensible cluster count and explain the next step
- 1Compute the percentage changes between consecutive coefficients: 11.8%, 13.0%, 16.5% and 67.9%.
- 1Identify the largest deterioration: merging from four clusters to three raises the coefficient by 67.9%.
- 1Stop before that poor merger, making four clusters the strongest candidate among these solutions.
- 1Fit K-means with K = 4 and compare stability, cluster sizes and nearby alternatives.
- 1Interpret original variable means and connect the profiles to reachable, sustainable and actionable decisions.
Key terms
- Customer Analytics
- The practice of deriving managerially relevant customer insight from data and translating it into a decision.
- Customer Relationship Management
- The strategic process of selecting customers a firm can profitably serve and shaping interactions with them.
- Customer Value
- The balance of benefits and costs for a customer, or the contribution a customer makes to the provider, depending on viewpoint.
- Customer Heterogeneity
- Differences in customer drivers, needs and behaviour that make a single marketing approach unsuitable for everyone.
- Sample Fitness
- The extent to which the observed customers and records can answer the stated research objective.
- Dummy Variable
- A binary indicator used to represent a category in a model, commonly with one category retained as the reference.
- RFM Segmentation
- A rule-based segmentation using recency, frequency and monetary value to group customers.
- Standardisation
- A transformation that expresses an observation relative to its variable mean and standard deviation.
- Hierarchical Clustering
- A distance method that progressively merges observations or groups and can help propose a cluster count.
- K-means Clustering
- A distance method that assigns observations to a chosen number of centres to reduce within-cluster distance.
- Latent Class Analysis
- A probabilistic method that explains binary response patterns through unobserved customer classes.
- Bayesian Information Criterion
- A comparative model-fit measure that rewards fit, penalises complexity and is preferred at lower values.
MARK5827 FAQ
How is the course assessed?
Four components total 100%: an Online Quiz worth 20%, a Case Study worth 40%, a Group Term Project worth 30%, and Peer Evaluation worth 10%. The first two are individual Moodle tasks. The project combines a tutorial presentation with a written report and requires coding.
How difficult is the course?
The challenge comes from joining three perspectives: customer relationship management, empirical analysis and coding. The teaching emphasises fundamentals rather than advanced statistics, but strong work must still choose suitable data, justify a method, interpret output, recognise limits and communicate a managerial action clearly.
How much R coding is required?
R is used in tutorials to apply customer analytics methods, and coding is required for the Group Term Project. The individual Moodle tasks state that coding is not required. The course treats programming as a means to customer insight, with emphasis on understanding and reproducible basic solutions.
What is the main analytics workflow?
Begin with the business objective and success criterion, identify and prepare relevant customer data, select variables and a sample, choose an analytical approach, interpret the result, deliver it to the decision process and learn from the outcome. Method choice should follow the decision rather than lead it.
How do clustering and latent classes differ?
Hierarchical and K-means clustering use distances among observed variables, while latent class analysis models hidden classes that make observed binary responses more or less likely. Clustering often uses an elbow rule to propose group count; LCA compares candidate models with BIC and interprets response probabilities.
How should I prepare for the group project?
Keep a decision log covering the objective, sample, variable definitions, cleaning choices, method, validation, interpretation and recommendation. Reproduce the analysis in a clean R session, reconcile every table and figure with the text, then translate the result into a clear executive action with limitations and measurement.
Questions students ask in each MARK5827 chapter
- Why should the business question come before the method?
- The objective determines which population, variables, time window and evidence are relevant. Choosing a method first encourages analysts to interpret whatever pattern appears, even when it cannot answer the manager's decision or support an action the organisation can implement. Chapter 01 · Customer Analytics for Managerial Decisions
- Why are there two views of customer value?
- Customers evaluate benefits such as function, convenience, emotion or symbolism against costs and risks. Firms evaluate lifetime, referral, knowledge and influencer contributions. An initiative can improve one side while harming the other, so a responsible analysis states both viewpoints. Chapter 02 · CRM, Customer Value and Privacy
- Can cleaning fix an unrepresentative sample?
- No. Cleaning can correct or document recorded values, but it cannot restore customers who were excluded by the selection rule. If the sample covers only premium or active customers, redesign the extraction before making claims about all customers or inactive groups. Chapter 03 · Data Preparation and Measurement
- When is RFM useful?
- RFM is useful when recent purchase, purchase frequency and monetary value capture behaviour relevant to the decision. It is transparent and quick, but its cut-offs are chosen by the analyst and a dimension may be weak in contexts where customers buy a product only once. Chapter 04 · RFM and Two-Step Cluster Analysis
- How is latent class analysis different from K-means?
- K-means forms groups by reducing distances between observed metric values and chosen centres. Latent class analysis estimates hidden groups that make binary responses more or less probable. LCA uses comparative model fit such as BIC, while K-means requires a chosen cluster count. Chapter 05 · Latent Class Analysis and Managerial Action
How to prepare for the assessments
Study the course as one decision process rather than a collection of software commands. First, practise translating a CRM concern into a research objective and a success criterion. Second, audit whether the sample and variables represent that objective before thinking about the method.
Third, learn each method through its decision role: RFM supplies transparent rules, hierarchical clustering proposes a group count, K-means fits centres, and latent class analysis describes hidden probability profiles. Recompute short quantitative decisions by hand, including K minus one dummy variables, percentage jumps in agglomeration coefficients, standardisation and the minimum BIC comparison.
After every output, write three sentences: what the pattern describes, what it does not establish, and what a manager should test or do next. Use R scripts to reproduce the analysis, but explain objects and results in plain language. For the group project, rehearse both audiences.
Preserve enough detail for an analyst to judge quality, then build a concise executive explanation around the business problem, the evidence, the recommendation, the limitation and the measurement plan. Review the official dates and task instructions in Moodle before submission.
Your AI Marketing tutor for MARK5827
Stuck on a hard MARK5827 question? Sia is AskSia’s AI Marketing tutor — ask any MARK5827 Customer Analytics question and get a clear, step-by-step explanation grounded in how the course is actually taught and assessed. Read this whole study guide free, then take your hardest questions to Sia.
