Segmentation, Targeting and Customer Profiling

Introduction and Context

Markets are rarely homogeneous. Customers differ in what they need, how they evaluate alternatives, how much they can spend, where they shop, and how they respond to communication. Segmentation, targeting and positioning (STP) provides a disciplined way to manage this diversity. Rather than treating every potential buyer as identical, the marketer identifies groups with meaningful similarities, evaluates their commercial and strategic value, selects priority groups, and designs a distinctive value proposition for them.

Segmentation is not simply dividing a database into age bands or geographic areas. It is an analytical and strategic process. A useful segment should be identifiable, sufficiently substantial, reachable, responsive to a distinctive marketing approach, and compatible with the organisation’s capabilities and brand purpose. Customer profiling then translates segment-level evidence into an interpretable description of typical customers, their contexts, motivations and barriers.

The process is iterative. Market evidence may reveal that an apparently attractive segment is too expensive to reach, that a demographic segment contains several different need states, or that a behaviour-based segment cuts across conventional age and income categories. Effective brand managers therefore combine quantitative analysis with qualitative understanding.

Learning Objectives

By the end of this chapter, students should be able to:

1. Explain the role of segmentation, targeting and positioning in brand management.
2. Apply demographic, geographic, psychographic, behavioural and needs-based segmentation.
3. Use cross-tabulation and association measures to interpret customer data.
4. Explain the purpose and limitations of clustering for market segmentation.
5. Develop evidence-based customer personas.
6. Assess segment attractiveness and organisational fit.
7. Convert segmentation findings into targeting and positioning decisions.
8. Recognise ethical, legal and practical limitations in customer profiling.

1. The STP Logic

The STP framework links market analysis to brand strategy:

Segmentation identifies groups that differ in relevant characteristics or responses.

Targeting evaluates these groups and selects one or more as strategic priorities.

Positioning defines how the brand should be understood by the chosen target relative to alternatives.

These activities are related but not interchangeable. A segment is an analytical group; a target is a managerial choice; a position is the intended meaning of the brand in the target customer’s mind.

A brand may use:

• Undifferentiated marketing, offering one broad proposition to the whole market.
• Differentiated marketing, developing distinct propositions for several segments.
• Concentrated marketing, focusing resources on one or a small number of segments.
• Micromarketing or personalised marketing, adapting offers to local areas, communities or individuals.

The appropriate approach depends on resources, market heterogeneity, competitive intensity, data quality and the risks of excessive fragmentation. Personalisation is not automatically superior: highly customised activity can increase cost, create inconsistent brand meaning and raise privacy concerns.

2. Bases of Segmentation

2.1 Geographic segmentation

Geographic segmentation groups customers by location, such as country, region, climate, urban density or neighbourhood type. It is useful when distribution, climate, language, regulation or local culture affects demand. A sportswear brand, for example, may distinguish customers in tropical and cold climates. Geography alone, however, is often a weak proxy for motivation; people living in the same city may have very different lifestyles and needs.

2.2 Demographic segmentation

Demographic variables include age, household composition, gender identity, income, occupation, education and life stage. They are relatively easy to measure and often useful for media planning, forecasting and product design. For example, household life stage may affect demand for convenience foods or financial services.

Demographics should not be treated as explanations in themselves. Age may correlate with media habits, but it does not prove that age causes a preference. When possible, demographic variables should be connected to observed behaviour or stated needs.

2.3 Psychographic segmentation

Psychographics concern values, attitudes, personality, interests, activities and lifestyle. Two customers with identical income and age may differ substantially in environmental concern, status orientation, risk tolerance or desire for novelty. Psychographic data may be collected through surveys, interviews, social listening and validated attitude scales.

Because psychographic constructs are often latent—that is, not directly observable—measurement quality matters. Questions should be clear, scales should be reliable, and the researcher should avoid assuming that a single response represents a stable personality trait.

2.4 Behavioural segmentation

Behavioural segmentation uses what customers do, including purchase frequency, recency, monetary value, category usage, channel preference, response to promotions, loyalty, product usage and churn risk. It often has strong managerial relevance because behaviour is closer to commercial outcomes than broad demographic descriptors.

A common descriptive framework is RFM analysis:

• Recency: how recently a customer purchased.
• Frequency: how often the customer purchased during a defined period.
• Monetary value: how much the customer spent during that period.

RFM scores are usually relative rankings rather than direct measures of future value. A high-spending customer is not necessarily profitable if service costs and returns are high.

2.5 Needs-based segmentation

Needs-based segmentation groups customers according to the problem they are trying to solve or the benefit they seek. Examples include convenience, reassurance, performance, belonging, affordability or self-expression. This approach is especially valuable for brand positioning because it links customer insight directly to value creation.

Needs-based segments may cut across demographics. A commuter, a student and a retiree may all seek a fast, reliable mobile payment service. Conversely, customers in the same demographic group may seek different benefits. Needs therefore require investigation through qualitative research, survey items, usage occasions and choice data.

2.6 Occasion and contextual segmentation

Demand can change by occasion: weekday versus weekend, routine versus celebration, individual versus group use, or urgent versus planned purchase. Contextual variables such as time, location, device and weather can make a customer’s need state dynamic. A person may belong to several practically relevant segments depending on the occasion.

3. Requirements of a Useful Segment

A segment should satisfy several criteria. No criterion is absolute; the assessment is strategic and context-dependent.

A segment can be large but unattractive if competition is intense or margins are low. A small segment can be valuable if it has high profitability, strategic influence or strong growth potential.

4. Cross-Tabulation for Segment Discovery

Cross-tabulation, or a contingency table, compares the distribution of one categorical variable across the categories of another. It is a foundational exploratory technique. For example, a brand manager may compare preferred channel by age group, or trial by need state.

Suppose a survey of 200 customers records preferred shopping channel:

Percentages should normally be calculated within rows when comparing channel preferences across age groups. Thus, 40 of 80 customers aged 18–29 prefer the app, or 50%. Among customers aged 50+, only 2 of 40, or 5%, prefer the app. The row denominator is essential; percentages based on the total sample answer a different question.

For a two-variable table, the expected count in cell i,j under independence is:

where Eᵢⱼ is the expected count, Rᵢ is the total for row i, Cⱼ is the total for column j, and N is the grand total. The chi-square statistic is:

Oᵢⱼ is the observed count. This test evaluates whether the variables are statistically independent in the sampled population, subject to assumptions such as independent observations and sufficiently large expected counts. Statistical significance does not establish managerial importance or causation. With a very large sample, a trivial difference may be statistically significant.

For a 3 × 3 table, degrees of freedom are:

where r is the number of rows and c is the number of columns. Analysts should also report effect size, practical magnitude and confidence intervals where appropriate. A significant association may be difficult to exploit if the difference is small or the groups cannot be reached efficiently.

5. Clustering and Data-Driven Segmentation

Clustering algorithms group observations so that members within a cluster are more similar to one another than to members of other clusters. Unlike a cross-tabulation, clustering can use multiple variables simultaneously. Common approaches include k-means, hierarchical clustering and model-based clustering.

5.1 k-means clustering

In k-means, the analyst selects a number k of clusters. The algorithm assigns each observation to the nearest centroid, recalculates the centroids, and repeats until assignments stabilise or improvement becomes negligible. For p measured variables, the usual objective is to minimise within-cluster squared Euclidean distance:

Here, Cₖ is cluster k, xᵢ is observation i represented as a vector of variables, and μₖ is the centroid of cluster k. The method assumes that numerical variables are meaningfully comparable after preparation and works best for relatively compact, roughly spherical clusters. It is sensitive to scaling and outliers.

Before clustering, analysts should inspect missing data, transform skewed variables where justified, standardise variables measured on different scales, and remove or treat extreme observations carefully. Standardisation commonly uses:

where x is an observation, μ is the sample mean and σ is the sample standard deviation. The resulting value expresses distance from the mean in standard-deviation units. Standardisation is appropriate when variables should receive comparable influence; it is not automatically appropriate when business importance should determine weighting.

Choosing k should combine statistical diagnostics, stability, interpretability and managerial usefulness. The elbow method examines diminishing reductions in within-cluster variation; silhouette analysis compares cohesion and separation. Neither method proves that a cluster is a real market segment. A cluster is useful only if it is meaningful, reachable and actionable.

5.2 Interpreting clusters

Clusters should be profiled after formation using variables not necessarily used to create them, such as profitability, channel response or satisfaction. Analysts should avoid naming clusters with unsupported stereotypes. “Mobile-first convenience seekers” is more informative than “young people” if the evidence concerns app usage and convenience needs.

Clustering is exploratory rather than self-validating. Different variable selections, starting points or sample periods may produce different solutions. Validation with a holdout sample, a later wave of data or qualitative research increases confidence.

6. Customer Profiling and Persona Development

A persona is a research-based, deliberately simplified representation of a priority customer type. It is not a fictional character created from intuition. A strong persona includes:

• a concise identity and contextual description;
• relevant demographic and geographic facts;
• goals, needs and desired benefits;
• behaviours, occasions and channel preferences;
• barriers, anxieties and switching triggers;
• decision criteria and information sources;
• a representative quotation grounded in research;
• implications for product, service, communication and measurement.

A persona should distinguish evidence from inference. For example, “62% of interviewed users abandon checkout when delivery timing is unclear” is evidence; “Alex hates online shopping” is an unjustified generalisation. Personas should also record the date, market and evidence base because customer behaviour changes.

Personas are useful communication devices. They help creative, product, sales and service teams imagine a coherent customer context. They become harmful when they replace data, conceal segment diversity or encourage discriminatory assumptions. A persona should support decisions, not determine eligibility for essential services without appropriate safeguards.

7. Assessing Segment Attractiveness

Segment evaluation combines market attractiveness with the organisation’s ability to compete. Useful dimensions include:

• current and forecast size;
• growth rate and expected demand;
• willingness to pay and margin potential;
• acquisition, service and retention costs;
• competitive intensity;
• accessibility through channels and media;
• strategic fit and brand credibility;
• risk, regulation and reputational implications;
• likely responsiveness to the proposed value proposition.

A simple weighted score can structure discussion:

Sⱼ is the score for segment j, wᵢ is the weight assigned to criterion i, and aᵢⱼ is the segment’s rating on criterion i. Weights should sum to 1 when expressed as proportions. Ratings must use a clearly defined scale, such as 1 to 5. The score is a decision aid, not an objective truth. Sensitivity analysis should test whether a small change in weights reverses the ranking.

For example, suppose a segment receives ratings of 4 for growth, 5 for margin, 3 for accessibility and 2 for competitive intensity. With weights of 0.30, 0.30, 0.20 and 0.20, its weighted score is:

S = (0.30 × 4) + (0.30 × 5) + (0.20 × 3) + (0.20 × 2) = 3.8

This calculation assumes that all ratings are oriented in the same direction, so a higher competitive-intensity rating must mean a more favourable competitive condition—or, alternatively, competition must first be reverse-coded. Mixing directions produces an invalid score.

8. From Target Selection to Positioning

Targeting is a choice about resource allocation. The organisation may prioritise one segment, several segments with related needs, or a sequence of segments over time. The decision should consider not only segment value but also the credibility of the brand’s proposed promise.

Positioning answers four questions:

1. For whom is the brand relevant?
2. What category or frame of reference does it belong to?
3. What distinctive benefit or meaning does it offer?
4. Why should customers believe the claim?

A useful positioning statement is an internal strategic tool, not necessarily advertising copy:

“For [target customer], [brand] is the [frame of reference] that [distinctive benefit] because [reason to believe].”

Positioning must be differentiated, valuable, credible, communicable and deliverable. A claim such as “the fastest service” requires operational evidence. If the brand cannot consistently deliver the promise, communication may create trial but damage trust and retention.

Segmentation should therefore influence the marketing mix. Product features, service design, pricing, distribution, media selection and brand language should reflect the chosen customer’s needs and context. Targeting is incomplete when it changes only advertising while the actual customer experience remains generic.

9. Illustrative Case: Repositioning a Refillable Beverage Brand

Background

The fictional brand ClearSpring sells refillable water bottles and filtration accessories through retail stores and an online channel. Sales have grown slowly despite rising public concern about single-use plastic. Management initially targeted “environmentally conscious consumers aged 18–35.”

Problem

The demographic definition was broad and produced weak campaign response. Research was required to identify actionable segments and determine whether sustainability alone was a sufficient positioning basis.

Analysis

The team combined three data sources: a 1,200-person survey, transaction records and 24 semi-structured interviews. The survey measured usage occasion, attitudes, convenience concerns, purchase frequency and willingness to pay. Transaction records supplied channel and repeat-purchase behaviour.

Cross-tabulation showed that high environmental concern was common across age groups, but willingness to carry a bottle daily was strongly associated with commuting and gym usage. Interviews identified three recurring need states:

1. Routine commuters wanted leak prevention, rapid refilling and a bottle that fitted a work bag.
2. Performance users wanted temperature retention, capacity and durable construction.
3. Occasional eco-buyers liked the sustainability message but considered refillable products inconvenient and expensive.

A clustering analysis using standardised measures of usage frequency, convenience importance, durability importance and sustainability attitude produced three interpretable groups broadly consistent with these need states. The team tested the solution on a later sample and found similar profiles, although cluster boundaries were not exact.

Approach and solution

The company evaluated the segments on size, repeat-purchase potential, willingness to pay, reachability, competitive intensity and fit with existing capabilities. Routine commuters were selected as the primary target because they showed frequent usage, strong dissatisfaction with leakage, and good reach through workplace and transit media. Performance users became a secondary target. Occasional eco-buyers were retained as a broader communication audience but were not given the highest acquisition budget.

The primary persona was “Maya,” a 29-year-old urban professional who commutes by train, carries a laptop bag, refills twice daily and dislikes cleaning difficult bottle components. The persona was explicitly based on survey and interview evidence rather than assumed lifestyle traits.

The positioning direction became: “For urban commuters who refill throughout the day, ClearSpring is the dependable refillable bottle that makes everyday hydration easy because its leak-resistant design, cleanable components and workplace refill support are engineered around daily routines.” Sustainability remained a supporting benefit rather than the sole promise.

The marketing programme introduced a leak-resistant lid, simplified cleaning instructions, commuter-focused demonstrations and partnerships with workplaces. Measurement included conversion rate, repeat purchase, return rate, customer satisfaction with leakage and the proportion of customers using the product at least four days per week.

Outcome

After six months, the pilot market showed higher trial-to-repeat conversion and fewer product returns than the previous broad campaign. The result cannot be attributed to targeting alone because product design and distribution also changed. Nevertheless, the case demonstrates how need-based insight can produce a more actionable target and a more credible position than a broad demographic label.

Learning points

The case illustrates five principles: demographic categories may conceal need differences; behavioural and attitudinal data are complementary; clusters require interpretation and validation; personas should be evidence-based; and positioning must be supported by the actual customer experience.

10. Ethics, Governance and Limitations

Customer profiling creates responsibilities. Data should be collected for legitimate purposes, with appropriate transparency, security and retention controls. Marketers should follow applicable privacy and data-protection law, including requirements concerning consent, lawful processing, access, deletion and automated decision-making where relevant.

Important risks include:

• using sensitive attributes or proxies in ways that exclude or disadvantage people;
• inferring health, financial vulnerability or other sensitive conditions without proper justification;
• combining datasets beyond the purpose for which data were collected;
• treating algorithmic clusters as natural or permanent identities;
• targeting vulnerable consumers with manipulative or exploitative messages;
• presenting correlation as causation;
• allowing inaccurate customer records to shape decisions.

Fairness assessment should examine outcomes across relevant groups, while respecting privacy and legal requirements. Governance should specify who may access data, how models are documented, how errors are corrected and when human review is required. An ethical segmentation strategy asks not only “Can this group be identified?” but also “Should it be targeted in this way?”

Technologies such as machine learning, real-time analytics and customer data platforms can improve identification and activation. They can also amplify biased historical data, encourage excessive surveillance and obscure decision logic. Human judgement, documentation, testing and ongoing monitoring remain necessary.

Best-Practice Workflow

A robust segmentation project can follow this sequence:

1. Define the strategic decision and business problem.
2. Specify the market and unit of analysis.
3. Combine relevant first-party, research and contextual data lawfully.
4. Clean, document and assess data quality.
5. Explore distributions, cross-tabs and meaningful associations.
6. Develop and validate segments using suitable analytical methods.
7. Profile segments with needs, behaviours, value and accessibility.
8. Evaluate attractiveness, fit, risks and sensitivity to assumptions.
9. Select target segments and define a positioning direction.
10. Translate the decision into the marketing mix and customer experience.
11. Test, measure and revise as market conditions change.

A technically sophisticated segmentation that does not change a decision is not successful analytics. The final test is whether the insight improves allocation, customer value and brand coherence.

Chapter Summary

Segmentation divides a heterogeneous market into meaningful groups; targeting selects priority groups; and positioning defines the desired meaning of the brand for those groups. Demographic, geographic, psychographic, behavioural, needs-based and occasion-based approaches answer different questions and are often strongest when combined.

Cross-tabulation reveals associations between categorical variables, but analysts must select appropriate denominators and distinguish statistical significance from managerial importance. Clustering can uncover multidimensional patterns, yet it depends on variable selection, scaling, algorithmic assumptions and validation. Personas convert evidence into a practical representation of a priority customer, but they must not replace research or encourage stereotypes.

Segment attractiveness involves size, growth, profitability, accessibility, competition, strategic fit, responsiveness and risk. Weighted scoring can structure judgement when criteria, weights and rating directions are explicit. Targeting decisions should lead to a credible, differentiated and deliverable positioning, supported by the complete customer experience. Ethical governance, privacy protection, fairness and ongoing monitoring are essential parts of professional customer profiling.

Key Terms

Segmentation: Dividing a market into groups with relevant similarities.

Targeting: Selecting market segments for strategic attention and resource allocation.

Positioning: Defining the intended meaning of a brand relative to alternatives.

Demographic segmentation: Grouping by characteristics such as age, income or life stage.

Psychographic segmentation: Grouping by values, attitudes, interests or lifestyles.

Behavioural segmentation: Grouping by observed or reported actions and usage patterns.

Needs-based segmentation: Grouping by problems, motivations or benefits sought.

Occasion segmentation: Grouping according to the context or event in which demand occurs.

Cross-tabulation: A table showing the joint distribution of two categorical variables.

Expected count: The count predicted in a table cell if the variables are independent.

Chi-square test: A test of association between categorical variables under stated assumptions.

Clustering: An analytical method that groups observations according to similarity.

Centroid: The representative mean point of a cluster in a numerical feature space.

Standardisation: Rescaling a variable using its mean and standard deviation.

Persona: An evidence-based representation of a priority customer type.

Segment attractiveness: The commercial and strategic appeal of a market segment.

Responsiveness: The extent to which a segment reacts differently to a tailored offer.

Value proposition: The benefits and value promised to a target customer.

Customer data platform: A system that organises customer data for analysis and activation.

Review Questions

Short Answer Questions

1. Distinguish segmentation, targeting and positioning.
2. Why can demographic variables be insufficient for explaining customer behaviour?
3. What is the purpose of a cross-tabulation?
4. State two limitations of k-means clustering.
5. Identify four components of an evidence-based persona.

Long Answer Questions

1. Compare demographic, psychographic, behavioural and needs-based segmentation, including an appropriate application for each.
2. Explain how a marketer should evaluate segment attractiveness and organisational fit.
3. Describe the process of preparing, conducting and interpreting a clustering analysis.
4. Explain how segmentation insights should influence positioning and the marketing mix.
5. Discuss ethical and governance risks associated with algorithmic customer profiling.

Analytical / Application Questions

1. A retailer finds that app usage differs substantially across age groups. Design a cross-tabulation analysis, specify the appropriate percentages, and explain what additional evidence is needed before targeting an age group.
2. Develop three potential segments for a meal-delivery brand using behavioural and needs-based variables. Propose a method for evaluating their attractiveness and justify a target choice.
3. Create an evidence-based persona and positioning statement for one segment identified in a university service of your choice. State which claims are evidence-based and which require further research.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes a target segment from a market segment?

A. A target segment is selected for strategic attention, whereas a market segment is an analytical group.

B. A target segment must always be larger than a market segment.

C. A target segment is defined only by demographics.

D. A target segment is the same as a brand position.

Correct Answer: A. A target segment is selected for strategic attention, whereas a market segment is an analytical group.

Question 2

When comparing channel preferences across age groups in a cross-tabulation, which denominator is usually most appropriate?

A. The grand total for the entire sample

B. The total within each age-group row

C. The largest cell in the table

D. The total number of channels

Correct Answer: B. The total within each age-group row

Question 3

What does a chi-square test of independence primarily assess?

A. Whether two categorical variables are statistically associated

B. Whether one brand has the highest profit

C. Whether a cluster has exactly equal sizes

D. Whether a persona is psychologically realistic

Correct Answer: A. Whether two categorical variables are statistically associated

Question 4

Why should variables commonly be standardised before distance-based clustering?

A. To make every customer identical

B. To prevent variables with larger measurement scales from dominating distance calculations

C. To guarantee that the chosen number of clusters is correct

D. To eliminate the need for validation

Correct Answer: B. To prevent variables with larger measurement scales from dominating distance calculations

Question 5

Which is a key limitation of k-means clustering?

A. It cannot use numerical variables

B. It is sensitive to scaling, outliers and the selected number of clusters

C. It always produces causal explanations

D. It can only analyse two variables

Correct Answer: B. It is sensitive to scaling, outliers and the selected number of clusters

Question 6

Which characteristic most directly indicates that a segment is actionable?

A. It has a memorable name

B. It can be reached and served with a distinctive marketing approach

C. It contains customers from only one age group

D. It is the smallest segment

Correct Answer: B. It can be reached and served with a distinctive marketing approach

Question 7

A weighted segment score uses weights 0.4, 0.3 and 0.3, with ratings 5, 3 and 4. What is the score?

A. 3.4

B. 3.8

C. 4.1

D. 4.7

Correct Answer: B. 3.8

Question 8

Which statement best describes a research-based persona?

A. A stereotype created from the manager’s intuition

B. A fictional customer with no connection to data

C. A simplified representation grounded in customer evidence

D. A legal classification of all customers

Correct Answer: C. A simplified representation grounded in customer evidence

Question 9

What is the main purpose of a positioning statement?

A. To list every feature of a product

B. To define the target, frame of reference, distinctive benefit and reason to believe

C. To calculate the sample size for a survey

D. To replace product and service delivery

Correct Answer: B. To define the target, frame of reference, distinctive benefit and reason to believe

Question 10

Why is statistical significance alone insufficient for a targeting decision?

A. It proves that the relationship is causal

B. A statistically significant difference may be too small or inaccessible to exploit commercially

C. It can only be calculated for qualitative data

D. It guarantees that a segment will be profitable

Correct Answer: B. A statistically significant difference may be too small or inaccessible to exploit commercially