Digital Research, Segmentation and Value Proposition

Introduction

Digital marketing decisions should be based on evidence rather than intuition alone. Organisations now have access to search queries, website behaviour, customer records, social conversations, platform analytics, reviews and direct responses from customers. These data sources can reveal what people need, how they evaluate alternatives, which messages influence them and where the customer experience fails. However, data do not automatically become insight. Effective digital marketers must frame research questions, select appropriate methods, assess data quality, interpret patterns cautiously and convert findings into actionable segmentation and value propositions.

This chapter examines the relationship between digital research, audience segmentation, targeting and value proposition design. It treats these activities as an integrated process: research identifies meaningful differences among customers; segmentation organises those differences into usable groups; targeting determines which groups the organisation will serve; and the value proposition explains why a chosen audience should prefer the organisation’s offering.

Learning Objectives

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

1. Distinguish primary, secondary, behavioural, social-listening and search data.
2. Formulate digital research questions and select suitable research methods.
3. Evaluate data quality, sampling, validity, reliability and ethical risks.
4. Use descriptive metrics and qualitative evidence to generate customer insights.
5. Construct useful digital audience segments using relevant variables.
6. Evaluate segment attractiveness and select priority target audiences.
7. Develop differentiated, evidence-based and testable customer value propositions.
8. Apply research and segmentation principles to a digital marketing case.

1. From Data to Marketing Insight

A useful distinction exists between data, information, insight and action. Data are recorded observations, such as 12,000 website visits, a review comment or a search term. Information is organised data that answers a basic question, such as the fact that mobile visitors have a lower checkout completion rate than desktop visitors. Insight is an interpreted explanation with strategic significance—for example, mobile shoppers may abandon because the payment process is difficult to use. Action is the decision taken in response, such as simplifying mobile checkout and testing the change.

A digital research process normally follows these stages:

1. Define the decision or management problem.
2. Translate the problem into research questions.
3. Identify relevant data sources and methods.
4. Collect, clean and document the data.
5. Analyse patterns and competing explanations.
6. Generate and prioritise insights.
7. Translate insights into segments, targeting choices and propositions.
8. Test outcomes and revise the decision.

A common failure is beginning with available data rather than with a decision. A dashboard may contain hundreds of metrics but still fail to explain why customers are leaving, which audience should be prioritised or what message should be tested.

Research Questions

A good research question is specific, decision-relevant and answerable with available evidence. “How can we improve digital performance?” is too broad. More useful questions include:

• Which customer groups have the highest likelihood of repeat purchase?
• What barriers prevent first-time visitors from completing checkout?
• Which benefits do environmentally conscious consumers associate with the brand?
• How does the brand’s perceived value differ from that of competitors?
• Which audience and message should be prioritised for a new campaign?

Research questions may be exploratory, descriptive, explanatory or evaluative. Exploratory research investigates an insufficiently understood problem. Descriptive research measures characteristics or behaviour. Explanatory research examines possible causes or relationships. Evaluative research assesses whether an intervention produced a desired result.

2. Digital Data Sources and Research Methods

Primary Research

Primary research is data collected specifically for the current research purpose. Common methods include surveys, interviews, focus groups, usability tests, diary studies, experiments and customer observation.

Surveys are useful for measuring attitudes, awareness, satisfaction, intentions and self-reported behaviour across relatively large samples. Good survey design requires clear wording, logical ordering, appropriate response scales and avoidance of leading or double-barrelled questions. “How satisfied are you with price and delivery?” is double-barrelled because a respondent may have different opinions about price and delivery. These should be measured separately.

Interviews provide depth and allow researchers to probe motivations, meanings and experiences. They are especially useful when the problem is poorly understood. Their limitations include interviewer influence, small samples and difficulty generalising findings.

Focus groups reveal how people discuss a category and respond to competing ideas. Group interaction can stimulate useful insights, but dominant participants, social conformity and moderator effects may distort responses.

Usability testing asks participants to complete realistic tasks on a website or application. It can identify navigation, comprehension and interaction problems that analytics alone cannot explain.

Experiments and A/B tests compare outcomes under controlled conditions. For example, one group may see a “Free delivery” headline while another sees “Fast delivery in two days.” Random assignment strengthens causal interpretation, although results can be affected by sample size, seasonality, novelty and exposure differences.

Secondary Research

Secondary research uses information collected previously by the organisation or by external parties. Internal sources include customer relationship management records, transaction data, web analytics, email performance and customer-service logs. External sources include government statistics, industry reports, academic studies, competitor communications and syndicated research.

Secondary data are often faster and less expensive than primary research. They can establish market size, category trends and historical performance. Nevertheless, marketers should check the purpose for which the data were originally collected, definitions of variables, time period, population coverage, methodology and possible commercial bias.

A market report may define “active customer” differently from a company’s CRM system. A competitor’s public engagement figures may not be comparable with the organisation’s private conversion data. Comparability must be established before conclusions are drawn.

Behavioural and First-Party Data

Behavioural data record actions such as page views, searches, product views, add-to-cart events, purchases, repeat visits, email clicks and app activity. First-party data are collected directly by an organisation through its own customer interactions, subject to applicable law and consent requirements.

Behavioural data are valuable because they capture observed actions rather than only stated intentions. They are not complete explanations, however. A customer may view a product because of curiosity, research a purchase offline or share a device with another person. Event tracking can also be incomplete or incorrectly configured. Behavioural evidence should therefore be interpreted alongside customer feedback and contextual information.

Social Listening

Social listening systematically monitors publicly available digital conversations, including posts, comments, reviews, forums and creator content. It can identify emerging needs, complaints, competitor comparisons, language used by customers and changes in sentiment.

Social listening is more than counting mentions. A high volume of negative conversation may result from one unusual event, while a small number of detailed complaints may reveal a serious product problem. Effective analysis considers:

• Volume: how frequently the topic appears.
• Reach: the potential audience exposed to the conversation.
• Engagement: reactions, comments and shares.
• Sentiment: positive, negative, neutral or mixed evaluation.
• Emotion: such as frustration, excitement or anxiety.
• Topic and context: what the conversation is actually about.
• Author influence and credibility: who is speaking and why.

Automated sentiment classification can misread sarcasm, slang, dialect, irony and context. Human review of a sample is important, particularly when decisions affect reputation or vulnerable groups. Public availability does not necessarily mean that every post may ethically be repurposed in a targeted campaign.

Search Data

Search data show what people actively seek, including product categories, problems, brands, locations and comparisons. Search volume can indicate interest, while query wording can reveal the consumer’s stage in the decision process. For example, “best running shoes for flat feet” expresses a need and problem context; “Brand X return policy” indicates post-purchase information seeking.

Search data should not be treated as a direct measure of demand without qualification. A query may be informational rather than commercial, and search volume can be affected by news, seasonality, media coverage or automated activity. Useful analyses include query classification, seasonality, geographic differences, device differences, related questions and the distinction between branded and non-branded searches.

Combining Methods

A strong research design often uses triangulation, the comparison of findings from multiple sources or methods. For example, analytics may show that users abandon a product page, interviews may reveal uncertainty about sizing, and search data may show frequent queries about fit. When these sources converge, confidence in the insight increases. When they disagree, the disagreement is itself a research question rather than a reason to select the most convenient result.

3. Measurement, Sampling and Data Quality

Descriptive Metrics

Digital marketing uses ratios to make performance comparable across different periods or audience sizes. A conversion rate is commonly defined as:

The denominator must be specified. A visitor-based rate, session-based rate and click-based rate answer different questions. If 240 of 8,000 eligible sessions result in a purchase, the session conversion rate is:

This calculation describes performance; it does not establish why the rate is 3% or whether a campaign caused it.

For a sample proportion, an approximate standard error is:

Here, p̂ is the observed sample proportion expressed as a decimal, n is the sample size, and the formula assumes a random or approximately representative sample with independent observations. It is most appropriate when observations are not clustered and the sample is sufficiently large for a normal approximation. For a conversion estimate of 0.03 based on 8,000 independent sessions, the approximate standard error is √(0.03 × 0.97 ÷ 8,000) ≈ 0.00191, or about 0.191 percentage points. This describes sampling uncertainty only; tracking error and selection bias may be larger.

A percentage change is:

If conversion rises from 3.0% to 3.6%, the relative percentage change is (3.6 − 3.0) ÷ 3.0 × 100 = 20%. The absolute change is 0.6 percentage points. These should not be confused.

Sampling

A population is the complete group of interest; a sample is the subset studied. Probability sampling gives members a known, non-zero chance of selection and supports stronger generalisation. Non-probability methods—such as convenience, quota, purposive and opt-in samples—can be practical but may produce selection bias.

A large sample is not automatically representative. An online survey with 50,000 voluntary respondents may systematically exclude people with limited access, low interest or different language preferences. Marketers should report who was included, who was excluded, how recruitment occurred and how missing data were handled.

Reliability and Validity

Reliability concerns consistency of measurement. A reliable satisfaction scale should produce reasonably consistent results when the underlying attitude has not changed. Validity concerns whether the measure captures the intended construct. A click is a reliable behavioural event if tracking is stable, but it may not validly measure purchase intention.

Threats include poorly worded questions, tracking changes, duplicate users, bot traffic, non-response, recall error, social desirability bias and confounding variables. Data documentation should record definitions, collection dates, filters, transformations and known limitations.

4. From Insight to Segmentation

Market segmentation divides a broad market into relatively distinct groups whose members share relevant characteristics, needs or responses. Segmentation is not simply describing demographic differences. A useful segment should support a different marketing decision.

Segmentation Bases

Demographic variables can be useful proxies, but they should not be assumed to explain motivation. Two people of the same age may have different needs, and people with different ages may seek the same benefit. Needs, behaviour and context are often more actionable.

Requirements for Effective Segments

Effective segments are:

• Measurable: their size, characteristics and outcomes can be estimated.
• Substantial: they are sufficiently valuable to justify attention.
• Accessible: the organisation can reach them through lawful and practical channels.
• Differentiable: they respond differently to an offer, message or channel.
• Actionable: the organisation can design and deliver a suitable programme.
• Stable enough: they persist long enough to support decisions, while remaining responsive to change.

A segment is not automatically useful because it is statistically discoverable. A cluster of customers may differ on many variables but be impossible to reach or economically unattractive.

Segmentation Approaches

A priori segmentation uses preselected rules, such as “customers who purchased in the last 90 days.” It is transparent and easy to activate. Post hoc segmentation discovers groups from observed data, using techniques such as clustering or latent-class analysis. It can reveal patterns that managers did not anticipate but requires careful interpretation and validation.

A practical segmentation workflow is:

1. Define the market, decision and unit of analysis.
2. Select variables linked to needs, behaviour or value.
3. Clean data and address missing or extreme values.
4. Explore distributions and relationships.
5. Create segments using rules or analytical methods.
6. Profile each segment with independent variables.
7. Test whether segments differ in meaningful outcomes.
8. Estimate reach, value, costs and risks.
9. Select target segments and design differentiated actions.
10. Monitor migration and update the model.

Clustering methods can be useful, but the number of clusters should not be chosen solely because software produces a result. Managers should assess interpretability, stability across samples, meaningful differences and operational feasibility. A technically elegant model that produces indistinguishable campaigns has little marketing value.

Personas

A persona is a research-based representation of a meaningful audience. It may include goals, context, barriers, behaviours, preferred information sources and decision criteria. A persona should not be a fictional stereotype. It should be traceable to evidence and should communicate a segment’s strategic relevance to creative, media, product and service teams.

5. Targeting and Prioritisation

Targeting is the selection of segments to serve. Possible strategies include:

• Undifferentiated targeting: one broad offer for the market.
• Differentiated targeting: distinct offers or messages for several segments.
• Concentrated targeting: focus on one or a few segments.
• Personalised or individual targeting: tailor communication or experiences at a person or account level, within ethical and legal limits.

Segment attractiveness should be evaluated alongside organisational fit. Relevant criteria include current size, growth, purchase potential, competition, cost to serve, strategic fit, data availability, brand compatibility and risk. A small segment may be attractive if it has strong margins and strategic importance; a large segment may be unattractive if acquisition costs are excessive.

Targeting decisions should also consider fairness. Algorithms may optimise response by excluding people with low predicted profitability, even when they have legitimate access needs. Sensitive attributes and their proxies require particular caution. Efficient targeting is not necessarily responsible targeting.

6. Customer Value Propositions

A value proposition is a clear statement of the value an organisation promises to create for a specified customer in a particular context. It explains the customer problem or desired outcome, the offered solution, the distinctive benefit and the reasons to believe the claim.

A useful structure is:

For [target customer] who [need or problem], [offering] provides [primary benefit] because [reason to believe], unlike [relevant alternative], which [limitation].

A value proposition is not the same as a slogan. A slogan may be memorable, while a value proposition must guide product design, content, channel selection and measurement. It should be customer-centred, specific, credible, differentiated and testable.

Value, Benefits and Costs

Customer value is multidimensional. Customers may value functional performance, convenience, emotional reassurance, social identity, status, risk reduction or sustainability. They also incur monetary, time, effort, psychological, privacy and switching costs.

A simplified conceptual representation is:

This is not a universal accounting identity and should not be interpreted as a directly measurable financial equation in every context. It is a framework for analysing trade-offs. A premium service may create greater functional and emotional benefits while also imposing a higher price. The proposition must make the benefits credible and relevant to the target audience.

Differentiation and Positioning

Differentiation concerns meaningful differences in the offering or experience. Positioning concerns the place the offering occupies in the target customer’s mind relative to alternatives. Possible bases include performance, convenience, expertise, trust, design, service, community, price or a distinctive combination.

A differentiated proposition should answer four questions:

1. Who is the priority customer?
2. What important job, problem or aspiration is addressed?
3. What benefit is promised?
4. Why should the customer believe the organisation can deliver it better or differently?

Claims must be supported by evidence such as product performance, independent certification, guarantees, reviews, transparent processes or demonstrated service levels. Vague claims such as “the best,” “innovative” or “personalised” are weak unless their meaning and proof are clear.

Testability

A proposition is testable when it produces observable predictions. For example, a hypothesis may be: “Among first-time mobile visitors seeking rapid delivery, a message emphasising guaranteed two-day delivery will produce a higher qualified checkout rate than a message emphasising low price.”

Testing requires a defined audience, comparison condition, outcome metric, exposure period and decision rule. The primary outcome should reflect the objective; clicks alone may be insufficient if they do not lead to profitable or satisfactory customers. Tests should avoid changing multiple major elements at once unless the purpose is explicitly multivariate experimentation.

7. Applied Case: GreenBasket Online Grocery

The following is an illustrative case.

Background

GreenBasket is an online grocery retailer operating in three metropolitan areas. Its traffic has grown, but first-order conversion and repeat purchase are below management expectations. The company markets organic and lower-packaging products but competes with supermarkets offering lower prices and faster delivery.

Problem

Management initially proposed a single campaign claiming “Healthy, sustainable groceries delivered affordably.” Research was required to determine whether the market contained distinct audiences and which proposition should be tested.

Analysis

The research team combined four sources:

• Website analytics for 180 days, including device, product category, delivery-slot selection and repeat purchase.
• A survey of 620 recent visitors measuring priorities, barriers and attitudes.
• Twenty semi-structured interviews with customers and non-customers.
• Social listening and search-query analysis concerning packaging, price, delivery and product quality.

Analytics showed that mobile users formed the majority of visits but completed checkout less often than desktop users. Interviews suggested that some mobile users found substitutions and delivery fees unclear. Search data showed two distinct patterns: queries about affordable weekly shopping and queries about plastic reduction and product provenance. Social discussions contained positive reactions to local sourcing but complaints about unpredictable substitutions.

The team identified three provisional segments:

1. Convenience planners: time-poor households seeking reliable weekly delivery and easy reordering.
2. Value-conscious families: shoppers highly concerned with total basket cost, promotions and predictable fees.
3. Purpose-led shoppers: customers willing to pay somewhat more for verified sourcing, reduced packaging and transparent producer information.

The segments were validated by comparing survey responses, purchase behaviour and interview accounts. They were not treated as absolute categories: customers could move between segments depending on household circumstances and shopping occasion.

Approach and Solution

GreenBasket selected convenience planners and purpose-led shoppers as initial targets. Value-conscious families were retained as a secondary audience because competing primarily on price would weaken margins and place GreenBasket against larger retailers.

Two propositions were developed:

• Convenience planners: “For busy households who want grocery shopping completed without repeated searching, GreenBasket provides dependable weekly delivery and simple reordering, with clear substitution choices and delivery fees shown before payment.”
• Purpose-led shoppers: “For shoppers who want lower-impact groceries without uncertainty about sourcing, GreenBasket provides verified local and lower-packaging options with transparent product and producer information.”

The team redesigned mobile checkout to show delivery fees and substitution preferences earlier. It then planned controlled tests comparing the new propositions with the generic sustainability-and-affordability message. Measures included qualified checkout conversion, average order value, repeat purchase within 60 days, customer complaints and cancellation rates.

Outcome

In the initial pilot, the segment-specific messages generated stronger engagement than the generic message. The most important learning was not simply that one advertisement received more clicks. The improved proposition aligned communication with operational changes: transparent substitutions and delivery information made the convenience claim more credible. Management therefore postponed broad scale-up until fulfilment reliability could meet the promise.

Learning Points

The case demonstrates that research sources are complementary; segmentation should be based on needs and behaviour rather than stereotypes; a target segment must be reachable and economically relevant; propositions require reasons to believe; and communication cannot compensate indefinitely for a poor customer experience.

8. Ethics, Privacy and Governance

Digital research creates responsibilities concerning consent, transparency, security, fairness and accountability. Applicable privacy law varies by jurisdiction, but good practice generally includes collecting only necessary data, stating purposes clearly, limiting retention, protecting access, respecting withdrawal choices and avoiding deceptive consent designs.

Researchers should distinguish anonymisation from pseudonymisation. Removing a name does not necessarily make data anonymous if individuals can be reidentified through combinations of location, time, device or behaviour. Data should be minimised and securely governed throughout its lifecycle.

Social listening requires respect for context and reasonable expectations. Sensitive conversations should not be exploited merely because they are publicly visible. Customer lists should not be combined with sensitive inferences without a lawful and ethical basis. Targeting systems should be audited for discriminatory outcomes, especially when automated models use proxies for protected characteristics.

AI tools can accelerate transcription, coding, clustering, copy generation and predictive modelling. They can also reproduce bias, invent patterns, expose confidential information or obscure how a decision was made. Human review, documented data provenance, validation against known cases and clear accountability remain necessary. AI should assist research judgement, not replace it.

9. Contemporary Developments and Best Practice

The decline of third-party identifiers and increasing privacy regulation are encouraging organisations to develop stronger first-party relationships, contextual advertising and consent-based data strategies. Customer data platforms can connect interactions, but integration does not guarantee accuracy or permission. Identity resolution should be conservative and transparent.

Search behaviour is also changing as people use visual search, voice interfaces, marketplaces and generative AI assistants. Marketers should focus less on isolated keywords and more on customer questions, useful content, structured information and trustworthy brand evidence. Social platforms increasingly combine commerce, entertainment and community, making creator credibility and comment quality important alongside reach.

Best practice includes:

• Begin with a decision, not a dashboard.
• Combine quantitative patterns with qualitative explanation.
• Define metrics and denominators before reporting performance.
• Check tracking quality and document assumptions.
• Avoid treating correlation as causation.
• Build segments that change decisions, not merely descriptions.
• Use needs, behaviour and context as well as demographics.
• Make propositions specific, credible and operationally deliverable.
• Test business outcomes, not vanity metrics alone.
• Review privacy, fairness and accessibility before activation.
• Reassess segments as needs, technology and markets change.

Chapter Summary

Digital marketing research converts evidence into decisions through a disciplined progression from problem definition to data collection, analysis, insight and testing. Primary research provides purpose-built evidence; secondary research supplies context; behavioural data reveal observed actions; social listening captures public conversation; and search data illuminate expressed information needs. Each source has strengths and limitations, so triangulation is essential.

Data quality depends on valid measurement, appropriate sampling, reliable tracking and careful interpretation. Metrics such as conversion rate and percentage change are useful only when their denominators and assumptions are explicit. Segmentation divides a market into groups that are measurable, substantial, accessible, differentiable and actionable. Needs-based and behavioural variables often provide stronger strategic guidance than demographic description alone.

Targeting selects priority groups by balancing attractiveness, organisational fit, reach, cost and ethical responsibility. A value proposition communicates the relevant customer problem, promised benefit, differentiation and reasons to believe. It must be supported by the actual customer experience and expressed as a testable hypothesis. Privacy, fairness, transparency and human accountability are integral to responsible digital research and targeting.

Key Terms

Primary research: Purpose-built collection of data for a current research problem.

Secondary research: Use of data previously collected for another purpose.

First-party data: Data collected directly by an organisation through its own interactions.

Behavioural data: Records of actions such as visits, clicks, purchases or searches.

Social listening: Systematic analysis of digital conversations and public feedback.

Search intent: The underlying purpose behind a search query.

Triangulation: Comparing multiple data sources or methods to strengthen interpretation.

Population: The complete group about which a study seeks to draw conclusions.

Sample: A subset of a population used for research.

Reliability: The consistency of a measurement.

Validity: The extent to which a measure captures its intended construct.

Market segmentation: Division of a market into groups with relevant shared characteristics.

Persona: An evidence-based representation of a target audience.

Targeting: Selection of segments for marketing attention and resource allocation.

Value proposition: A statement of the value promised to a specified customer.

Differentiation: Meaningful distinction between an offering and alternatives.

Positioning: The place an offering occupies in the customer’s comparative perception.

A/B test: Controlled comparison of two versions of an experience or message.

Conversion rate: The percentage of an eligible population completing a defined action.

Data minimisation: Collection and retention of only data necessary for a stated purpose.

Review Questions

Short Answer Questions

1. What is the difference between data, information and insight?
2. Identify two strengths and two limitations of social listening.
3. What makes a market segment actionable?
4. Why should a value proposition include a reason to believe?
5. Distinguish reliability from validity.

Long Answer Questions

1. Explain how primary, secondary, behavioural, social-listening and search data can be combined to investigate a digital marketing problem.
2. Discuss the main criteria used to evaluate and select target segments.
3. Explain why demographic segmentation alone may be insufficient for digital marketing decisions.
4. Describe the relationship between differentiation, positioning and value proposition design.
5. Evaluate the ethical and governance issues associated with personalised digital targeting.

Analytical / Application Questions

1. An online retailer has high mobile traffic but low mobile checkout completion. Design a research plan using at least three methods, and explain how each method would contribute different evidence.
2. Construct two differentiated value propositions for a streaming service serving students and retired customers. State the customer need, promised benefit, reason to believe and outcome metric for each.
3. A campaign’s conversion rate increases from 2.5% to 3.0%. Calculate the relative percentage change and explain why the result alone does not prove that the campaign caused the improvement.

Multiple Choice Questions with Answers

Question 1

Which sequence best represents the movement from raw evidence to managerial action?

A. Action → data → insight → information

B. Data → information → insight → action

C. Information → action → data → insight

D. Insight → data → action → information

Correct Answer: B. Data → information → insight → action

Question 2

Which research method is most suitable for identifying why users abandon a confusing mobile checkout process?

A. A review of national population statistics only

B. A usability test combined with interviews

C. A demographic census without website observation

D. A count of social media followers

Correct Answer: B. A usability test combined with interviews

Question 3

If 240 purchases occur in 8,000 eligible sessions, what is the session conversion rate?

A. 0.3%

B. 3%

C. 30%

D. 32%

Correct Answer: B. 3%

Question 4

Which statement about a large voluntary online sample is most accurate?

A. It is automatically representative because it is large

B. It eliminates non-response bias

C. It may still suffer from selection bias

D. It always supports causal conclusions

Correct Answer: C. It may still suffer from selection bias

Question 5

Which is a requirement of an effective market segment?

A. It must be defined only by age

B. It must be inaccessible to competitors

C. It must be actionable by the organisation

D. It must contain every potential customer

Correct Answer: C. It must be actionable by the organisation

Question 6

What is the main purpose of triangulation in digital research?

A. To replace all qualitative research with analytics

B. To compare evidence from multiple sources or methods

C. To increase advertising frequency automatically

D. To guarantee that all findings are causal

Correct Answer: B. To compare evidence from multiple sources or methods

Question 7

Which element makes a value proposition more credible?

A. A vague claim of being the best

B. A reason to believe supported by evidence

C. A list of every product feature

D. A promise unrelated to customer needs

Correct Answer: B. A reason to believe supported by evidence

Question 8

A rise from 3.0% to 3.6% conversion represents what relative percentage change?

A. 0.6%

B. 6%

C. 20%

D. 60%

Correct Answer: C. 20%

Question 9

Which statement best distinguishes targeting from segmentation?

A. Segmentation selects media; targeting collects data

B. Segmentation divides the market; targeting selects groups to serve

C. Segmentation creates slogans; targeting measures validity

D. Segmentation is ethical review; targeting is data cleaning

Correct Answer: B. Segmentation divides the market; targeting selects groups to serve

Question 10

Why should automated social sentiment analysis receive human review?

A. Algorithms cannot count any digital activity

B. Public posts are always statistically invalid

C. Sarcasm, slang and context can be misclassified

D. Human review guarantees a causal result

Correct Answer: C. Sarcasm, slang and context can be misclassified