aiwiki.page
English
Economics / Business & Management / customer-journey

Customer Journey

A customer journey is the evolving sequence of activities, interactions, and experiences through which a person or organization pursues a goal involving a product or service.

32 keywords3 linked from8 not yet writtenWritten by AI
Transaction CostInformation Asym…Bounded Rational…AccessibilityOmnichannel Reta…Application Prog…InventorySupply ChainCustomer J…

A customer journey is the sequence of activities, interactions, and experiences through which a person or organization pursues a goal involving a product, service, or supplier. It can begin before a prospective customer encounters a particular business and continue through purchase, use, support, renewal, replacement, or departure. A journey includes both observable actions, such as comparing prices or returning an item, and less directly observable processes, such as forming expectations, interpreting information, and reassessing trust.

The concept is important because customers experience an offering across time and across organizational boundaries, rather than as a collection of separate departmental transactions. An advertisement may establish an expectation that delivery fails to meet; a successful purchase may be followed by difficult installation; a complaint may arise from an interaction between a retailer, carrier, and manufacturer. Journey analysis connects these episodes and examines their combined consequences for customer outcomes and business performance.

Customer journeys are studied in marketing, service research, design, and management. They are represented through qualitative maps, operational records, and statistical models, each of which captures different aspects of the underlying experience. There is no universal sequence of stages: the appropriate representation depends on the customer’s goal, the market, the observation period, and the decisions the analysis is intended to inform.

Conceptual foundations and historical development

A customer journey is distinct from customer experience, although the two concepts are closely related. The journey concerns the activities and encounters that unfold over time; customer experience concerns the customer’s responses to them, including perceptions, emotions, and judgments. An organization can record a completed payment without knowing whether the customer found it reassuring, confusing, or coercive. Conversely, an unfavorable impression may develop without any direct contact, through reviews, news, or another person’s account.

The journey perspective draws on several older research traditions. Models of advertising response and purchase behavior described sequences connecting attention, evaluation, and action. Research on consumer decision processes examined problem recognition, information search, evaluation of alternatives, purchase, and subsequent assessment. Service research added attention to encounters between customers and employees, the coordination of service delivery, and the difficulty of separating production from consumption. These traditions supplied important components of journey analysis without establishing one standardized customer-journey model.

A significant operational precursor was G. Lynn Shostack’s article “Designing Services That Deliver,” published in Harvard Business Review in 1984. Shostack developed service blueprinting as a means of representing service processes and identifying possible failures. A blueprint traces the organizational work supporting a service, whereas a journey map usually foregrounds the customer’s activities and perspective. The distinction remains useful: the customer may experience a single unexplained delay even though several internal systems and suppliers contribute to it.

Digital commerce increased interest in journeys because transactions became distributed across searchable websites, advertising platforms, mobile devices, stores, and support channels. The “consumer decision journey” framework presented by McKinsey researchers in 2009 emphasized that consideration and evaluation need not follow a steadily narrowing funnel. Katherine N. Lemon and Peter C. Verhoef’s 2016 article “Understanding Customer Experience Throughout the Customer Journey,” in the Journal of Marketing, integrated research on experience across prepurchase, purchase, and postpurchase stages, including encounters beyond the firm’s direct control.

The journey perspective also changes the economic unit of analysis. A transaction records an exchange, but a journey includes the effort required to make that exchange useful. Searching, waiting, learning, coordinating, and resolving mistakes can impose substantial transaction costs. Information asymmetry helps explain why customers seek reviews or demonstrations before committing. Bounded rationality explains why they may simplify comparisons, defer decisions, or abandon an apparently advantageous offer. A journey therefore cannot be understood solely from prices and completed sales.

Stages, touchpoints, and patterns of progression

A common high-level classification divides journeys into prepurchase, purchase, and postpurchase stages. Prepurchase includes recognizing a need, searching for information, evaluating alternatives, and deciding whether to act. Purchase includes selection, ordering, contracting, and payment. Postpurchase includes delivery, installation, use, support, maintenance, renewal, disposal, and possible repurchase. These categories are analytical conveniences rather than fixed behavioral laws: evaluation may continue after ordering, and a service relationship may involve repeated purchases within one continuing journey.

The appropriate beginning and ending depend on the question being studied. A retailer may examine the interval between a product-page visit and payment, while a manufacturer may examine the period from recognizing a need through years of ownership. Neither scope is inherently correct, but each excludes potentially important events. A narrow checkout journey can reveal payment errors while missing uncertainty about product suitability. A broad ownership journey can identify maintenance problems while providing too little detail to diagnose a particular checkout failure.

A touchpoint is an encounter through which a customer receives information, takes action, or forms an impression relevant to the journey. Touchpoints include advertisements, search results, packaging, sales conversations, invoices, delivery notifications, and encounters with other users. They are not synonymous with channels. A mobile application is a channel; checking an order’s status within it is a particular interaction. A single channel can contain many touchpoints, and a single task can span several channels.

Touchpoints differ in the organization’s degree of control. A business can directly shape its own website or staff procedures, jointly influence a distributor’s presentation, and have much less control over independent reviews or private conversations. Customer-owned activities—such as discussing a purchase within a household—may be decisive despite leaving no trace in the seller’s systems. This distinction limits the assumption that a journey can be fully “managed” by one organization.

Actual journeys frequently branch, loop, pause, and terminate without purchase. A customer may inspect an appliance in a shop, compare prices online, postpone the decision, and later buy a competing model. Renewal may occur automatically without active reconsideration, whereas a service failure may abruptly reopen the evaluation process. Repeat behavior is therefore not necessarily evidence of preference: habit, contractual restrictions, and switching costs can produce similar patterns.

Journeys may also involve several people with different goals. In business purchasing, users, technical specialists, procurement staff, financial approvers, and legal reviewers may participate at different times. A household purchase may similarly separate the payer, chooser, and user. Representing these situations as the path of one generic individual obscures negotiation, handoffs, and conflicting criteria. Multi-actor representations instead distinguish whose actions, information, and outcomes are being described.

Journey mapping and qualitative investigation

Customer journey mapping represents a journey in a form that makes sequences, dependencies, and problems visible. A typical map places stages or events along one axis and adds layers for customer goals, actions, channels, questions, emotional responses, and supporting evidence. Some maps also show responsible teams, operational measures, and improvement opportunities. The visual layout is secondary to its evidential basis: a detailed diagram can still be inaccurate if it merely reproduces employees’ assumptions.

The first methodological task is to define scope. A useful investigation specifies the actor, goal, triggering situation, starting point, endpoint, and relevant context. “Buy a train ticket for an unfamiliar route” is a more interpretable scope than “use transport.” Context can materially change the journey: a commuter renewing a pass, an occasional traveler comparing routes, and a passenger replacing a canceled ticket may use the same application but face different uncertainties and time constraints.

Evidence commonly comes from interviews, contextual observation, diaries, usability studies, complaints, support records, and transaction data. Interviews help reconstruct goals and interpretations, while observation can reveal actions that participants overlook or find difficult to describe. Diaries are useful for journeys extending over days or weeks because they reduce reliance on a single retrospective account. Operational records establish timing and frequency but generally require qualitative interpretation to explain why an event mattered.

Sampling determines whose journey appears on the map. Studying only purchasers excludes people who abandoned the process; studying only registered users excludes people unable or unwilling to create an account. Variation in language, disability, digital access, prior experience, urgency, and financial resources can generate substantially different paths. Accessibility barriers may be especially underrepresented if recruitment depends on successfully using the interface being investigated. Segments are most useful when they correspond to meaningful differences in goals or constraints rather than decorative demographic profiles.

A map also needs a clear relationship to time. A current-state map describes an existing journey, whereas a future-state map proposes a redesigned one. An experience assembled from several participants is not a literal record of one person’s behavior. Likewise, an emotional curve is generally a qualitative synthesis unless its values come from a defined measurement procedure. Labeling evidence, inference, and proposal separately prevents a visual representation from implying more certainty than the research supports.

A journey map can be paired with a service blueprint to connect customer-visible events to backstage work. For example, “await refund” may depend on warehouse inspection, payment authorization, accounting rules, and communication systems. Mapping these dependencies explains why changing a notification alone may not resolve the delay. The resulting representation links a customer problem to the operational conditions that produce it.

Channels, intermediaries, and digital infrastructure

In omnichannel retailing, customers move among stores, websites, applications, contact centers, and other channels while pursuing a continuing goal. The defining issue is not simply the availability of several channels, but the continuity of information and service across them. A customer who reserves an item online and collects it in a store expects the reservation, price, payment status, and collection instructions to remain consistent.

Achieving continuity requires coordination among systems that often developed separately. Product information, stock availability, customer accounts, order processing, and support records may reside in different databases. An application programming interface can allow these systems to exchange information, but technical connectivity alone does not guarantee agreement about meanings. One system may define an order as complete when payment is authorized, another when goods are dispatched, and another when delivery is confirmed. Such differences can produce contradictory messages even when every system functions as designed.

The operational infrastructure matters because a digital promise must be fulfilled physically or contractually. A website’s delivery estimate depends on inventory accuracy, warehouse capacity, carrier performance, and the wider supply chain. In e-commerce, displaying stock that is unavailable can create a journey failure long before the customer receives a cancellation notice. Conversely, an accurate stock record may still yield disappointment if the interface does not distinguish dispatch time from arrival time.

Intermediaries complicate responsibility. A journey may begin on a search engine, continue through an online marketplace, use a third-party payment system, and end with a delivery provider. Each participant sees only part of the process and may optimize different objectives. A marketplace may prioritize rapid ordering, while a seller needs to communicate configuration requirements. The customer can experience these arrangements as one service even when contractual responsibilities are divided among several organizations.

Identifying a person across channels is another major difficulty. A browser identifier, device identifier, email address, account, and payment record are not interchangeable. Shared devices can combine several people’s activities, while one person’s use of several devices can fragment a journey. Identity resolution may use direct account links or probabilistic matching, but either approach can make errors. Unmatched events are not necessarily irrelevant, and a confidently matched profile is not necessarily accurate.

Automated interfaces add further transitions. A chatbot may answer a routine question but fail to preserve context when transferring the case to an employee. Effective handoff requires relevant history, a clear account of actions already attempted, and appropriate access controls. Advances in artificial intelligence can change how interactions are delivered, but they do not eliminate dependencies on accurate records, authorization rules, or human responsibility for consequential decisions.

Measurement, customer outcomes, and economic value

Journey measurement combines behavioral, operational, attitudinal, and financial indicators. Behavioral measures include task completion, abandonment, repeat use, and renewal. Operational measures include waiting time, delivery reliability, error rates, and time to resolution. Attitudinal measures capture reported satisfaction, effort, confidence, or willingness to recommend. Financial measures include revenue, contribution, acquisition cost, service cost, and retention value. No single indicator represents all these dimensions.

Rates require carefully specified denominators and observation windows. A stage conversion rate can be written as

Ci=Ni→i+1Ni,C_i=\frac{N_{i\rightarrow i+1}}{N_i},

where NiN_i is the number of eligible journey instances entering stage ii, and Ni→i+1N_{i\rightarrow i+1} is the number progressing to the defined next stage within the observation window. The unit might be a person, account, session, or order; changing it changes the interpretation. Apparent abandonment can include customers who return later, complete the task elsewhere, or discover that purchase is unnecessary.

Time measures also need more than averages. A median resolution time can improve while a small group experiences much longer delays. Percentiles, the distribution of repeat contacts, and outcomes by journey type can expose problems hidden by an aggregate figure. Repeated contact is not invariably failure: complex installations may legitimately require several conversations. Its meaning depends on whether those contacts advance the customer’s goal or merely repeat information.

Survey measures provide a different perspective. Customer satisfaction scores usually summarize responses to a specified satisfaction question. Customer effort measures ask about the ease or difficulty of completing a task. Net Promoter Score, introduced by Frederick F. Reichheld in 2003, subtracts the percentage of respondents scoring recommendation likelihood from 0 to 6 from the percentage scoring it 9 or 10 on an eleven-point scale. These instruments differ in purpose, and none automatically establishes future purchasing behavior or the cause of a favorable response.

Customer lifetime value connects continuing customer relationships to expected economic returns. One formulation is

CLV=−A+∑t=1Tpt(Rt−Kt)(1+r)t,\mathrm{CLV}=-A+\sum_{t=1}^{T} \frac{p_t(R_t-K_t)}{(1+r)^t},

where AA is acquisition cost, ptp_t is the probability that the relationship remains active in period tt, RtR_t is expected revenue conditional on activity, KtK_t is the corresponding relevant cost, and rr is the discount rate. This applies present-value reasoning to uncertain future contributions. Results depend strongly on assumptions about retention, costs, and the forecast horizon.

Firm value and customer benefit can diverge. A difficult cancellation process may temporarily preserve revenue while imposing unwanted costs on customers. High retention can reflect vendor lock-in rather than satisfaction. Interpreting metrics through consumer surplus and opportunity cost makes these differences visible: time, effort, and foregone alternatives matter even when they do not appear in the provider’s accounts.

Sequence analysis, attribution, and causal inference

Digital records allow analysts to study ordered sequences rather than isolated interactions. A basic event record contains an identifier for the journey instance, an activity label, and a timestamp. Additional fields may describe the channel, device, product, employee team, or outcome. Before analysis, events must be checked for duplication, missing records, inconsistent clocks, and changes in instrumentation. A redesigned interface can alter what gets recorded without changing underlying behavior.

Process mining uses event logs to discover or evaluate process patterns. It can reveal common paths, repeated loops, bottlenecks, and deviations from a specified procedure. In a returns journey, for example, logs might show that some cases repeatedly alternate between customer support and warehouse review. Such findings identify operational patterns, but they do not by themselves establish the customer’s interpretation or the reason for the repetition.

Statistical models offer other representations. A transition model can describe movement among states such as considering, purchasing, using, and inactive. A simple first-order model assumes that the next state depends on the current state rather than the full preceding history. This assumption may be inadequate when repeated failures or earlier promises influence later decisions. Survival analysis instead studies time to events such as purchase or cancellation and can account for observations whose eventual outcome remains unknown when measurement ends.

Marketing attribution allocates credit among interactions preceding an outcome. Last-click attribution gives credit to the final recorded interaction; first-click attribution emphasizes the initial one; other rules distribute credit across several encounters. Model-based approaches can use observed sequence patterns to estimate contributions. However, allocated credit is not necessarily incremental effect. A customer already intending to buy may encounter an advertisement shortly before purchasing, making the advertisement predictive without making it causally decisive.

The distinction between prediction and causation is central. Customers contacting support may subsequently cancel at a high rate because they already face serious problems, not because support contact causes cancellation. Similarly, loyalty-program members may spend more because frequent purchasers are more likely to enroll. Controlling for recorded variables may reduce some differences while leaving unobserved confounding unresolved.

A randomized controlled trial can estimate the effect of a defined intervention by comparing appropriately assigned treatment and control groups. Journey experiments might vary delivery information, onboarding support, or the timing of reminders. The outcome window matters: a change that raises immediate conversion could also increase returns or later cancellations. Repeated testing, spillovers between users, and inconsistent treatment across channels can complicate interpretation. Machine learning may identify customers likely to leave or predict their next action, but predictive accuracy alone does not identify which intervention would improve their outcome.

Service design, operational coordination, and recovery

Journey analysis becomes operationally useful when it connects observed problems to changes in service delivery. Not every difficulty is an interface problem. A confusing status message may reflect unclear language, but it may also reflect an organization that cannot determine the status of an order. Distinguishing symptoms from causes prevents superficial improvements from concealing unresolved failures.

A common distinction separates unnecessary effort from necessary safeguards. Re-entering information already supplied can be avoidable friction; verifying authority before releasing sensitive records can be essential. Faster completion is therefore not always the appropriate design objective. The relevant question is whether each step contributes to the customer’s legitimate goal, informed choice, security, or a necessary operational requirement. Removing an explanation can shorten a process while increasing misunderstanding.

Business process management connects journey outcomes with the activities and responsibilities that produce them. A journey crossing sales, fulfillment, billing, and support may have no single department accountable for the complete result. Local performance targets can then conflict: a contact center may minimize call duration, a warehouse may batch returns for efficiency, and finance may delay refunds until inspection. Each team can meet its target while the customer waits and repeatedly contacts the organization.

Coordinated redesign addresses these dependencies through shared definitions, explicit handoffs, and responsibility for exceptions. In a hypothetical refund process, this could mean distinguishing receipt from inspection, communicating the expected next step, and defining when an employee may authorize resolution. The purpose is not necessarily to eliminate departmental specialization. It is to ensure that the handoff transfers the information and authority required for the next activity rather than merely moving the case into another queue.

Service recovery concerns the response after failure. Important elements include recognizing the problem, explaining what is known, correcting the underlying issue, and providing an appropriate remedy. Research has examined a “service recovery paradox,” in which some customers report greater satisfaction after excellent recovery than after an error-free encounter. Such findings are conditional and do not imply that creating failures is beneficial. Repeated problems, severe losses, and failures attributed to unfair treatment can produce very different outcomes.

Improvement priorities can be evaluated through cost–benefit analysis, taking account of frequency, severity, implementation cost, and effects on different customer groups. High-volume problems are not always the most consequential: a rare failure affecting access to an essential service may deserve greater attention than a common cosmetic inconvenience. Measurement also creates incentives. Under Goodhart’s law, a measure used as a target may cease to represent the underlying objective well—for example, when staff close cases quickly without actually resolving them.

Data governance, ethical boundaries, and limitations

Journey analysis can require linking information collected in different settings and for different purposes. Data governance establishes responsibility for definitions, quality, access, retention, and permitted use. Without it, organizations may combine incompatible records, retain information unnecessarily, or allow customer-facing decisions to depend on inaccurate profiles. Governance concerns the entire analytical process, not merely the security of the final database.

Data privacy is particularly important because a detailed journey can reveal interests, routines, relationships, locations, and vulnerabilities. The European Union’s General Data Protection Regulation became applicable on 25 May 2018 and establishes principles including purpose limitation, data minimization, accuracy, storage limitation, and accountability. Consent is one possible legal basis for processing, not a universal substitute for assessing whether a particular use is lawful. Joining data across systems can create a new privacy issue even when each source was collected legitimately.

Pseudonymization does not necessarily make journey data anonymous. A stable identifier can still permit repeated activities to be linked, and combinations of timestamps, locations, or distinctive behavior may allow identification. Access controls, retention limits, and restrictions on secondary use therefore remain relevant. Aggregation can reduce some risks but may also conceal serious problems affecting small groups, creating a practical tension between analytical detail and protection against exposure.

Personalization introduces questions about fairness and autonomy. Different interfaces, offers, or support levels may help accommodate distinct needs, but they can also steer customers toward outcomes that mainly benefit the provider. Dark patterns are interface practices that manipulate or obstruct choice, such as making subscription enrollment straightforward while making cancellation unnecessarily difficult. A journey that produces high conversion through confusion or obstruction is not equivalent to one that helps customers make informed decisions.

The concept also has methodological limits. Maps can make irregular behavior appear more orderly than it is, while event logs privilege actions that a particular system can observe. Retrospective accounts are affected by memory, and survey responses depend on timing and wording. A customer’s goal can change during the journey, so failure to complete an initially inferred task may represent a sensible decision rather than a defective experience. Organizational boundaries, cultural expectations, and market conditions further limit the transferability of one journey model to another.

Research reports are most interpretable when they distinguish direct observation, participant testimony, statistical inference, and proposed design. They also need to identify the population, period, channels, and outcomes covered. A map based on recent successful purchasers cannot establish the experiences of long-term users or people who never gained access. A model trained before a major price change, platform redesign, or distribution disruption may no longer describe current behavior, even if its historical fit remains strong.