Channels, Customer Relationships and Growth Systems

Introduction

A business model becomes commercially real only when a venture can reach customers, deliver its value proposition, establish a relationship that customers accept, and generate economically sustainable growth. In the Lean Canvas, these activities are often compressed into the channels and customer segments fields. In practice, however, channel design is a system that links market discovery, acquisition, sales, delivery, onboarding, usage, retention, payment, service, and referral.

A channel is not merely a communications medium. It is the route through which a customer encounters an offer, evaluates it, purchases or adopts it, receives value, and potentially recommends it. A customer relationship is not merely customer service. It is the designed pattern of interaction, trust, support, personalization, and participation that connects the customer to the venture over time.

For graduate entrepreneurs and managers, the central question is therefore not “Which channel should we use?” but:

Which combination of channels and relationship mechanisms enables this particular customer segment to perceive, obtain, adopt, use, and repeatedly choose the value proposition at an acceptable cost and risk?

This chapter examines physical, digital, and omnichannel models; distinguishes acquisition from distribution and retention; introduces growth systems and referral loops; and provides tools for assessing product–channel fit.

Learning Objectives

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

1. Distinguish channels, touchpoints, customer relationships, and growth systems.
2. Design acquisition, sales, distribution, onboarding, retention, and referral pathways.
3. Compare physical, digital, and omnichannel channel models.
4. Apply customer-journey and funnel logic to identify bottlenecks.
5. Assess product–channel fit using qualitative and quantitative evidence.
6. Evaluate channel economics, including conversion, retention, customer acquisition cost, and contribution margin.
7. Address channel conflict, privacy, accessibility, ethical, regulatory, and governance concerns.
8. Develop a testable channel strategy appropriate to a Lean Canvas.

1. The Channel as a Business-System Choice

1.1 Core terminology

Channel: A route or mechanism through which a venture communicates with, sells to, delivers value to, and supports a customer.

Touchpoint: A specific interaction between a customer and the venture, such as an advertisement, sales call, product page, store visit, delivery, or support exchange.

Customer relationship: The pattern and degree of interaction a venture maintains with a customer before, during, and after exchange.

Growth system: A connected set of mechanisms through which customer acquisition, activation, retention, revenue, and referral reinforce or constrain one another.

Product–channel fit: The degree to which a product’s characteristics, customer segment, buying process, and channel economics are mutually compatible.

A channel can be owned, partnered, or earned. An owned channel is controlled by the venture, such as its website, sales force, application, or retail outlet. A partnered channel uses another organization’s reach, such as distributors, marketplaces, affiliates, consultants, or franchisees. An earned channel arises when customers, communities, media, or algorithms generate attention without a direct payment for every exposure.

Channel decisions have strategic consequences. They influence:

• who can access the product;
• how much trust is required before purchase;
• the speed of learning from customers;
• the cost and predictability of acquisition;
• delivery reliability and service quality;
• data ownership and privacy exposure;
• margins, bargaining power, and scalability; and
• the type of relationship the venture can credibly promise.

A premium professional service, for example, may require consultative selling and high-touch onboarding. A low-cost mobile utility may depend on self-service discovery, rapid activation, and product-led referral. Treating both as “digital channels” would conceal important differences.

1.2 Channel functions across the customer lifecycle

A complete channel system performs several distinct functions:

These stages are analytically distinct even when one channel performs several functions. A marketplace may create awareness, enable conversion, and process payment, while the venture itself handles onboarding and support.

2. Customer Relationships: From Transactions to Designed Interaction

Customer relationships range from largely automated self-service to deeply relational account management. Common forms include:

• Self-service: Customers solve problems through product interfaces, documentation, configuration tools, or automated support.
• Automated service: Software, recommendation systems, chatbots, and triggered communications personalize the experience without continuous human intervention.
• Personal assistance: A representative supports the customer during purchase or use.
• Dedicated account management: A named manager coordinates a high-value or complex relationship.
• Community relationship: Customers receive value from interaction with one another, often moderated by the venture.
• Co-creation: Customers contribute ideas, content, data, testing, or design input.
• Transactional relationship: The interaction is intentionally limited to a discrete purchase and delivery.

No relationship type is universally superior. The appropriate design depends on customer risk, product complexity, frequency of use, switching costs, customer lifetime value, and the customer’s willingness to pay for assistance.

A useful design principle is relationship congruence: the promised relationship should match the customer’s job and the product’s risk. A hospital procurement team may expect implementation support, compliance documentation, and responsive account management. A consumer buying a low-cost digital template may prefer instant access and minimal human contact. Over-servicing reduces margin; under-servicing reduces trust and retention.

2.1 Relationship quality and trust

Trust is especially important when customers cannot fully evaluate quality before purchase. It may be built through:

• credible guarantees and transparent terms;
• demonstrations and independent evidence;
• secure payment and privacy safeguards;
• reliable fulfillment;
• visible expertise and social proof;
• responsive problem resolution; and
• consistency across touchpoints.

Trust is cumulative but fragile. A venture that acquires customers through exaggerated advertising but delivers slowly may achieve high initial conversion and poor retention. Channel metrics must therefore be interpreted across the entire relationship, not at the point of acquisition alone.

3. Physical, Digital and Omnichannel Models

3.1 Physical channels

Physical channels include retail stores, field sales, trade shows, distributors, clinics, warehouses, service centers, and face-to-face professional networks. Their strengths commonly include sensory demonstration, local trust, immediate human reassurance, and suitability for products requiring installation or physical inspection.

Their limitations include geographic constraints, staffing costs, inventory requirements, slower experimentation, and difficulty standardizing service. Physical channels may nevertheless be strategically superior where customers value touch, safety, status, consultation, or immediate possession.

3.2 Digital channels

Digital channels include websites, applications, search, social platforms, email, online marketplaces, webinars, digital communities, and software integrations. They can offer broad reach, rapid experimentation, automated service, detailed behavioral data, and low marginal distribution costs.

Digital channels also create risks: platform dependence, algorithmic volatility, cybersecurity threats, privacy obligations, digital exclusion, impersonality, and intense competition for attention. A digital channel is not automatically inexpensive; content production, paid media, software, support, fraud prevention, and data governance can be substantial costs.

3.3 Omnichannel systems

An omnichannel model coordinates channels so that the customer experiences one coherent relationship rather than disconnected encounters. A customer might discover a product through social media, examine it on a website, visit a store, purchase through a mobile device, receive delivery notifications, and obtain support through messaging. The channels differ, but identity, inventory, pricing, service history, and promises should be consistent where appropriate.

Omnichannel is not the same as multichannel. Multichannel means that several channels exist. Omnichannel means that the channels are deliberately integrated around the customer journey.

Integration can be operationally difficult. It requires common data definitions, inventory visibility, consent management, service standards, attribution rules, and organizational coordination. It may also create channel conflict if a direct online channel competes with distributors or physical retailers.

4. Designing the Customer Journey and Growth Funnel

A customer journey should be mapped from the customer’s perspective, not the organization’s departmental structure. For each stage, identify the customer’s goal, uncertainty, action, evidence required, venture intervention, and failure risk.

A simplified growth funnel may be represented as:

where N_t is the active customer base at time t, A_t is newly activated customers during the period, and C_t is customers who churn or become inactive. The identity is a stock-and-flow relationship: it applies when the active base is clearly defined and the flows are measured over the same period.

The funnel is not necessarily linear. Customers may enter through referrals, return after abandonment, compare several times, or move from a physical interaction to a digital purchase. A journey map should therefore represent loops and parallel paths rather than assuming one universal sequence.

4.1 Acquisition

Acquisition creates qualified attention or demand. Channels may include:

• search and educational content;
• paid advertising;
• outbound sales;
• partnerships and resellers;
• professional events;
• communities and social participation;
• referrals;
• public relations; and
• product features that expose the offer to new users.

The relevant question is not simply which channel produces the most leads. It is which channel produces customers who activate, retain, pay, and recommend at acceptable cost.

4.2 Sales and conversion

Sales design should reflect buying complexity. A low-risk, low-price product may use direct checkout. A complex enterprise solution may require qualification, diagnosis, business-case development, procurement, legal review, implementation planning, and executive sponsorship.

A venture should distinguish lead volume from qualified demand. A channel that produces many inexpensive inquiries can be inferior to one that produces fewer, high-probability opportunities. Sales stages should have explicit exit criteria, such as a validated problem, identified decision maker, agreed use case, budget pathway, and scheduled implementation.

4.3 Distribution and delivery

Distribution includes the physical or digital movement of the value proposition to the customer. Questions include:

• Who holds inventory or service capacity?
• Who controls fulfillment quality?
• How are returns, failures, and exceptions handled?
• Can delivery scale without degrading reliability?
• What information must be shared among partners?
• What service-level commitment is being made?

A channel is strategically weak if it generates demand that the delivery system cannot fulfill. Growth without operational capacity can damage reputation and increase support costs.

4.4 Onboarding and activation

Acquisition has little economic value if customers do not reach first meaningful value. Activation is the earliest observable behavior that indicates the customer has begun receiving the promised benefit. It must be defined specifically. For a collaboration application, it might be inviting a team and completing a shared task; for a financial service, it might be completing identity verification and making a first transaction.

Good onboarding reduces time to value, cognitive load, uncertainty, and setup errors. It may use guided configuration, templates, checklists, demonstrations, human assistance, or progressive disclosure. However, forced onboarding steps should not be added merely to increase data collection; they should help the customer achieve an outcome.

4.5 Retention and expansion

Retention results when the customer repeatedly receives sufficient value relative to price, effort, risk, and alternatives. Retention mechanisms include habit formation, reliable performance, integration into workflows, accumulated data, community belonging, responsive support, switching costs, and ongoing improvement.

Expansion can occur through higher usage, additional users, premium features, geographic extension, or complementary products. It should follow demonstrated value rather than exploit customer dependence.

4.6 Referral

Referral is a growth mechanism when a customer’s use makes another customer more likely to adopt. It may be explicit, as in a referral program, or organic, as in public advocacy and network visibility. Incentives require careful design: rewards should not encourage deceptive recommendations, spam, or unsuitable enrollment.

A referral loop can be described as:

1. A customer receives value.
2. The product creates a credible reason or mechanism to share.
3. A new prospect encounters the shared proof or invitation.
4. The prospect adopts and becomes a potential referrer.
5. The loop is measured for quality, not merely volume.

5. Channel Economics and Measurement

Quantitative metrics are useful only when definitions and time periods are consistent. Common measures include:

Conversion rate

The denominator must be specified. A visitor-to-purchase rate is not interchangeable with a lead-to-customer rate.

Customer acquisition cost

CAC should include the costs relevant to the selected boundary, such as paid media, commissions, sales labor, agency fees, and acquisition technology. The numerator and customer count must cover the same period and attribution rule.

Retention rate

For a cohort measured over a defined period:

The definition of “active” must be behaviorally meaningful. Login alone may not indicate value realization.

Contribution margin

Variable costs may include transaction fees, fulfillment, usage-based infrastructure, support that scales with usage, and partner commissions. Fixed costs should not be subtracted if the purpose is to assess unit-level contribution.

A simplified lifetime value estimate for a subscription with constant monthly revenue and constant monthly churn is:

This approximation assumes a stable monthly churn rate, stable contribution margin, no discounting, no expansion or contraction, and an indefinite geometric survival process. It is not appropriate when cohorts behave differently, contracts have fixed terms, margins change materially, or acquisition and retention costs are being confused. A more complete discounted model would sum expected future contribution margins by period.

Worked example

An online professional-learning service spends $24,000 on attributable acquisition and sales during one quarter and obtains 300 new paying customers. Its CAC is therefore:

Each customer generates $40 in monthly revenue. Variable service and payment costs are $16 per month, so monthly contribution margin is $24. If monthly churn is 5%, the simplified lifetime value estimate is:

The ratio of estimated LTV to CAC is 6:1. This result is not proof that the model is healthy: the estimate ignores discounting, support variation, expansion, implementation costs, and the time required to recover CAC. It does indicate that retention and margin deserve further investigation rather than immediate channel expansion.

5.1 Attribution limitations

Attribution assigns outcomes to channels, but customer journeys are often multi-touch. Last-click attribution may over-credit the final interaction and under-credit education, community, or brand-building. First-click attribution has the opposite weakness. Experiments, cohort analysis, incrementality tests, qualitative interviews, and triangulation are therefore preferable to a single dashboard number.

6. Assessing Product–Channel Fit

Product–channel fit is stronger when the channel’s natural operating logic supports the product’s adoption requirements. Assessment should examine six alignments:

1. Customer alignment: Does the channel reach the intended segment in a context where the problem is salient?
2. Trust alignment: Does the channel provide the credibility required for the purchase risk?
3. Economic alignment: Can expected margin support the channel’s acquisition, sales, fulfillment, and service costs?
4. Capability alignment: Can the venture deliver consistently through the channel?
5. Behavioral alignment: Does the channel match how customers search, compare, buy, use, and seek help?
6. Learning alignment: Does the channel provide sufficiently rapid and reliable feedback?

A channel hypothesis should be written in testable form: “For segment S, channel C will produce qualified customers who complete activation behavior A within time T at acquisition cost no greater than K, while achieving retention R after period P.” The exact thresholds are managerial assumptions, not universal laws; they should be justified by margins, cash constraints, strategic goals, and benchmark evidence.

Experiments should isolate a meaningful uncertainty. Examples include:

• a landing-page test of a narrowly defined value proposition;
• a small paid-search campaign compared with founder-led outreach;
• a distributor pilot in one territory;
• a concierge onboarding cohort;
• a referral invitation offered only after a customer reaches a value milestone; or
• an in-person demonstration followed by digital self-service.

The objective is not to maximize activity but to learn whether a repeatable pathway exists.

7. Applied Case: GreenRoute Workplace Meals

The following is an illustrative case, constructed for educational purposes.

Background

GreenRoute is a hypothetical venture offering scheduled, lower-waste workplace lunches from local kitchens. Its initial target segment is medium-sized firms with 100–500 employees. The value proposition combines convenient ordering, predictable delivery, dietary visibility, and reduced packaging waste.

Problem

The founders initially used social media advertising and a consumer-facing ordering website. Traffic was strong, but conversion was weak and repeat orders were lower than expected. Interviews revealed that employees liked the concept, yet workplace purchasing decisions were made by office managers. Employees also needed confidence that delivery would be coordinated and dietary information would be reliable.

Analysis

The original channel optimized awareness among individuals but ignored the buying unit and operational complexity. The product required organizational permission, scheduled delivery, kitchen coordination, and trust in allergen information. A purely self-service consumer funnel was therefore poorly aligned with the product and segment.

The team mapped the journey:

• Office manager becomes aware through a facilities-network event.
• Manager requests a workplace trial.
• GreenRoute conducts a short operational assessment.
• Employees receive a co-branded ordering link.
• Kitchens prepare a consolidated delivery.
• The manager receives a service report.
• Employees are invited to reorder or recommend the service to another team.

Approach and solution

GreenRoute adopted a hybrid model. It used partnership with workplace-services consultants for acquisition, founder-led sales for the first organizational contract, a digital employee ordering interface for conversion, and human implementation support for the first two deliveries. It also created a dietary-information protocol and a service recovery process.

The venture measured not only website conversion but also:

• qualified workplace opportunities;
• trial-to-contract conversion;
• employee activation, defined as a completed first order;
• on-time delivery;
• repeat order rate by workplace cohort;
• contribution margin after delivery and support costs; and
• referrals from managers and employees.

Outcome

The hybrid channel generated fewer raw leads than social advertising but produced more qualified opportunities and higher first-order completion. Onboarding costs were initially higher because of human assistance. After recurring operational patterns were documented, some onboarding tasks were automated. The venture retained human intervention for dietary exceptions, service failures, and complex workplace configurations.

Learning points

1. The user, buyer, and beneficiary may be different people.
2. A channel must fit the buying process, not merely the user’s media habits.
3. Human support can be a deliberate learning investment rather than an operational failure.
4. Digital delivery can coexist with physical distribution and relational sales.
5. Retention evidence should be measured at the cohort and account level.
6. A channel decision is inseparable from service design and operational capacity.

8. Governance, Ethics and Contemporary Developments

Channel systems collect and transmit personal, behavioral, payment, location, and sometimes sensitive data. Ventures should apply data minimization, purpose limitation, meaningful consent where required, access controls, retention policies, and clear explanations of automated decisions. Privacy compliance is jurisdiction-specific; organizations should obtain appropriate legal advice rather than treating a generic privacy notice as sufficient governance.

Accessibility is both an ethical and strategic requirement. Digital interfaces should support users with disabilities, while physical channels should consider mobility, sensory, language, and affordability barriers. Dark patterns—interfaces that manipulate users into unwanted consent, subscriptions, or sharing—damage autonomy and may violate consumer-protection rules.

Artificial intelligence can support recommendation, customer-service triage, demand forecasting, personalization, lead scoring, and content production. Its use requires attention to biased targeting, hallucinated service claims, opaque decisions, excessive surveillance, and the transfer of confidential data to external systems. Human escalation should remain available where errors could materially affect safety, rights, finances, or access.

Contemporary channel strategy is also shaped by platform concentration, privacy restrictions on third-party tracking, social commerce, creator-led distribution, conversational interfaces, embedded finance, and increasingly integrated physical–digital experiences. These developments create opportunities but also increase dependence on intermediaries. A prudent venture develops owned customer relationships and portable first-party data, while maintaining transparent consent and contractual safeguards with partners.

9. Managerial Best Practices and Common Failure Modes

Effective channel design usually follows these practices:

• Begin with the customer’s job, buying unit, risk, and context.
• Separate awareness, conversion, delivery, activation, retention, and referral metrics.
• Test one major channel uncertainty at a time where feasible.
• Measure cohorts rather than relying only on aggregate averages.
• Include fulfillment, support, returns, commissions, and implementation in economic analysis.
• Design service recovery before scaling acquisition.
• Keep a direct relationship with customers where strategic and legally appropriate.
• Establish channel ownership, data standards, and conflict-resolution rules.
• Match human assistance to complexity and customer value, not organizational habit.
• Treat referrals as an outcome of value and trust, not merely an incentive campaign.

Common failure modes include confusing reach with demand, optimizing clicks instead of retained customers, ignoring the economic buyer, adding channels before the first one is understood, promising omnichannel consistency without systems integration, and assuming that digital distribution eliminates service costs. Another frequent error is premature automation: a venture automates a poorly understood process and thereby scales confusion or failure.

Chapter Summary

Channels are designed pathways through which a venture creates awareness, enables evaluation and purchase, delivers value, supports use, retains customers, and stimulates referral. Customer relationships range from self-service and automation to personal assistance, account management, community, and co-creation. The appropriate relationship depends on product complexity, customer risk, expected value, and customer preferences.

Physical channels can provide trust, demonstration, and local access; digital channels can provide reach, speed, automation, and behavioral data; omnichannel systems coordinate several contexts but require substantial integration and governance. A complete growth system connects acquisition, sales, distribution, onboarding, activation, retention, expansion, and referral.

Metrics such as conversion rate, CAC, retention rate, contribution margin, and LTV can inform decisions when definitions, attribution boundaries, cohorts, and time periods are explicit. Simplified LTV formulas require strong assumptions and should not be treated as universal truths. Product–channel fit is assessed through customer, trust, economic, capability, behavioral, and learning alignment.

The central managerial discipline is to test a repeatable pathway from qualified attention to retained value. Growth should follow evidence of customer benefit and operational capability, while privacy, accessibility, fairness, security, and channel governance remain integral to responsible design.

Key Terms

Channel: A route through which a venture communicates, sells, delivers, or supports value.

Touchpoint: A specific interaction between a customer and a venture.

Customer relationship: The designed pattern of interaction maintained with a customer.

Growth system: Interconnected mechanisms for acquisition, activation, retention, revenue, and referral.

Owned channel: A channel controlled directly by the venture.

Partnered channel: A channel operated with or through another organization.

Earned channel: Attention or distribution generated by customers, communities, or media.

Physical channel: A channel involving tangible locations, people, or physical delivery.

Digital channel: A channel mediated primarily through digital technologies.

Omnichannel: Coordinated channel experiences organized around a continuous customer journey.

Activation: The point at which a customer first experiences meaningful value.

Retention: Continued customer activity or relationship over a defined period.

Churn: Customer loss or movement from an active to an inactive state.

Customer acquisition cost (CAC): Attributable acquisition and sales expenditure per new customer.

Contribution margin: Revenue less variable costs attributable to serving a customer.

Lifetime value (LTV): The expected contribution generated by a customer over the relationship.

Product–channel fit: Compatibility among product characteristics, customer behavior, channel capabilities, and economics.

Attribution: The process of assigning credit for an outcome to one or more touchpoints.

Channel conflict: Competition or inconsistency among channels serving the same market.

Dark pattern: An interface design that manipulates users toward choices contrary to their interests.

Review Questions

5 Short Answer Questions

1. Distinguish a channel from a touchpoint.
2. What is the difference between multichannel and omnichannel strategy?
3. Define activation and explain why it matters.
4. Identify two limitations of last-click attribution.
5. What does product–channel fit mean?

5 Long Answer Questions

1. Explain how acquisition, distribution, onboarding, retention, and referral form an integrated growth system.
2. Compare physical, digital, and omnichannel models using reach, trust, cost, learning speed, and risk.
3. Discuss how customer risk and product complexity should influence relationship design.
4. Explain the assumptions and limitations of the simplified LTV formula.
5. Analyze the governance and ethical issues associated with AI-enabled channel personalization.

3 Analytical / Application Questions

1. A graduate software venture has many website visitors but few users complete the first meaningful workflow. Design a diagnostic plan covering activation, onboarding, channel quality, and customer interviews.
2. A high-value industrial product is being sold through both a direct sales team and distributors. Propose an omnichannel governance system that addresses attribution, pricing, data sharing, service standards, and channel conflict.
3. Design a channel hypothesis and experiment for a subscription service. Specify the target segment, channel, activation behavior, economic metric, retention measure, decision threshold, and ethical safeguards.

Multiple Choice Questions with Answers

Question 1

Which statement best describes an omnichannel model?

A. It uses only digital channels to reduce operating costs.

B. It offers several channels that operate independently.

C. It coordinates multiple channels around a coherent customer journey.

D. It relies exclusively on physical sales representatives.

Correct Answer: C. It coordinates multiple channels around a coherent customer journey.

Question 2

What is activation in a customer journey?

A. The first time an advertisement is displayed

B. The point at which a customer first experiences meaningful value

C. The moment a company hires a sales representative

D. The point at which a customer unsubscribes

Correct Answer: B. The point at which a customer first experiences meaningful value

Question 3

A venture spends $18,000 on acquisition and gains 150 customers. What is its CAC?

A. $12

B. $120

C. $1,200

D. $2,700

Correct Answer: B. $120

Question 4

Which denominator is most appropriate for a lead-to-customer conversion rate?

A. Total employees in the venture

B. Qualified leads entering the defined sales process

C. Total website page views from all time periods

D. Total customers retained from previous years

Correct Answer: B. Qualified leads entering the defined sales process

Question 5

Which condition most strongly supports a high-touch relationship?

A. A low-risk product requiring no explanation

B. A complex, high-value product with significant adoption risk

C. A product purchased only through an automated vending machine

D. A product whose customers reject all assistance

Correct Answer: B. A complex, high-value product with significant adoption risk

Question 6

Using the simplified model, what is estimated LTV when monthly contribution margin is $30 and monthly churn is 10%?

A. $3

B. $30

C. $300

D. $3,000

Correct Answer: C. $300

Question 7

Which is a common limitation of last-click attribution?

A. It cannot record any digital transaction.

B. It tends to over-credit the final touchpoint and under-credit earlier influences.

C. It measures retention rather than acquisition.

D. It can be used only for physical retail.

Correct Answer: B. It tends to over-credit the final touchpoint and under-credit earlier influences.

Question 8

Which example best illustrates product–channel fit?

A. Selling a complex enterprise system only through an anonymous banner advertisement

B. Using consultative sales and implementation support for a complex organizational purchase

C. Choosing a channel solely because a competitor uses it

D. Selecting the channel with the highest number of impressions regardless of retention

Correct Answer: B. Using consultative sales and implementation support for a complex organizational purchase

Question 9

Which practice is most consistent with responsible AI-enabled channel personalization?

A. Collecting all available data without informing customers

B. Removing human escalation for high-impact decisions

C. Using data minimization, transparency, safeguards, and appropriate human review

D. Optimizing engagement even when recommendations are misleading

Correct Answer: C. Using data minimization, transparency, safeguards, and appropriate human review

Question 10

Why should a venture measure cohorts rather than relying only on aggregate retention?

A. Cohorts eliminate the need to define an active customer.

B. Cohorts reveal how customers acquired at different times behave over a comparable period.

C. Cohorts guarantee that attribution is causal.

D. Cohorts measure advertising impressions more accurately than customers.

Correct Answer: B. Cohorts reveal how customers acquired at different times behave over a comparable period.