RESEARCH DOSSIER
Consumer Motivation, Purchase Intentions, and Actual Behavior
Abstract. Consumer motivation, purchase intentions, and observed purchasing form a central research problem in consumer and economic psychology: how do goals become behavioral commitments and how, under actual market conditions, do those commitments become transactions? This fundamental critical narrative synthesis traces the development of intention-based behavioral theories, purchase forecasting, self-determination and habit accounts, constructive choice, implementation intentions, and field research on prices, information, and market behavior. It evaluates original experiments, panel studies, randomized interventions, clickstream analyses, and methodological reviews concerning behavioral prediction, self-generated survey validity, deferred decisions, and the intention–behavior gap. The evidence indicates that intentions carry meaningful but incomplete predictive information; specificity, purchase horizon, product familiarity, realized opportunity, pricing, and survey procedure affect the relationship. Experimental changes in expressed intentions need not yield comparable changes in objective behavior, while market interventions may affect acquisition through multiple causal pathways. The synthesis distinguishes desire, expectation, commitment, platform engagement, payment, and postpurchase outcomes; assesses measurement error and selection; and addresses heterogeneity, welfare, privacy, and delegated human–artificial purchasing. Its central conclusion is that consumer action emerges through dynamically interacting motivational, informational, material, and technological conditions. The work synthesizes published research and advances no claim of a newly established empirical mechanism. It is a critical narrative research monograph rather than an original experiment, preregistered systematic review, or meta-analysis.
Keywords: consumer motivation; purchase intention; purchase behavior; intention–behavior gap; theory of planned behavior; consumer decision-making; purchase conversion; consumer goals; self-generated validity; mere-measurement effect; behavioral intention; purchase forecasting; behavioral control; shopping cart abandonment; field experiments; digital consumer behavior; implementation intentions; consumer autonomy; consumer choice architecture; human–artificial decision assistance
Introduction: Consumer Action as a Multistage Psychological Process
Consumer motivation, purchase intention, and actual purchasing describe connected but empirically distinguishable events. Motivation concerns the psychological processes that make an acquisition goal desirable, urgent, meaningful, or worth pursuing. Purchase intention is an expressed commitment, plan, expectation, or judged likelihood of acquiring something under specified conditions. Actual behavior includes observable acts such as paying for a product, completing an order, initiating a subscription, renewing a contract, or abstaining from purchase. The research problem is how motivational states become expressed intentions and how, under changing opportunities and constraints, intentions become transactions. A reported preference, product-page visit, cart addition, and verified sale each occupy a different position in this sequence. Treating them as interchangeable would erase the very discrepancy that consumer psychology seeks to explain. Fishbein & Ajzen (1975) Morwitz (2014)
The present monograph examines the formation of consumer goals, the assessment of future acquisition, prospective planning, behavioral control, situational affordances, timing, and the conditions that produce completed or abandoned transactions. Its empirical unit is a specified purchase opportunity involving a person, an object or service, a time window, an accessible seller or channel, and an observable outcome. Psychological factors are studied alongside prices, income, inventory, trust, information, and institutional rules because each can alter the path from deliberation to action. The principal explanatory claim is methodological: a model of purchasing is scientifically meaningful only when it identifies both the mental constructs and the behavior it proposes to predict. The source corpus ranges from social-psychological theories and controlled choice experiments to marketing panels, field interventions, and digital transaction records. Ajzen (1991) Morwitz et al. (2007) DellaVigna (2009)
Three distinctions organize the inquiry. First, wanting a product need not imply wanting to pay its current price, within the available time, through an accessible channel. Second, what respondents report in an intention survey can be an estimate of future behavior rather than a settled plan; the wording and probability scale are therefore part of the measurement apparatus. Third, completed purchases cannot establish that a preceding intention caused the behavior, because product availability, habits, social cues, and promotional exposures may influence both. These distinctions demand designs that separate preference formation, intention measurement, opportunity, and purchase observation. The analysis is a critical narrative synthesis of published literature; it reports no new participant sample, preregistered systematic search, pooled effect estimate, or original causal experiment. Sheeran & Webb (2016) Chandon et al. (2005)
Historical Foundations: From Attitudes to Forecasts of Purchasing
Modern research on consumer intention grew at the intersection of attitude psychology, decision theory, and market forecasting. Fishbein and Ajzen's 1975 book, Belief, Attitude, Intention, and Behavior, formulated a systematic approach to behavioral prediction based on beliefs, evaluations, normative influences, and intentions. Its enduring methodological contribution was the insistence that the target action, context, and observation interval must be compatible across predictors and outcomes. Liking a brand is a general evaluation; intending to purchase a particular model next month is an action-specific forecast. A scientific study loses precision when it relates one to the other without matching their levels of specificity. These principles subsequently informed many consumer surveys, marketing forecasts, and studies of planned acquisition. Fishbein & Ajzen (1975) Ajzen (1991)
Ajzen's 1991 theory of planned behavior extended the analysis by considering perceived behavioral control, emphasizing that intended actions often require resources, ability, and opportunity. A consumer might evaluate an electric bicycle favorably, perceive social approval, and sincerely intend to buy it while doubting whether the requisite financing or storage space will be available. The theory predicts that attitudes, subjective norms, and perceived control contribute to intentions and that intentions and control perceptions are associated with later behavior. Its value is as a specified, testable model: the theory identifies distinct predictors, their expected paths, and the circumstances under which intention should be a better guide to action. The exact explanatory contribution of perceived control requires behavioral measurement rather than relying solely on another self-report. Ajzen (1991) Armitage & Conner (2001)
The marketing tradition directly investigated the forecasting quality of purchase questions. Morwitz, Johnson, and Schmittlein found in 1993 that asking respondents about purchase intention could itself affect later buying; intention surveys could therefore participate in the phenomenon they ostensibly observed. The methodological issue became sharper in Chandon, Morwitz, and Reinartz's 2005 analysis of self-generated validity: the relationship between stated intention and subsequent action can be exaggerated by the process of measurement. Later comparisons of varied markets identified conditions under which purchase intentions align more closely with sales, including shorter horizons, familiar products, and more specific targets. By the 2010s, the field had converged on a practical proposition: intention measures carry information, but their predictive value is conditional and must be calibrated against behavior. Morwitz et al. (1993) Chandon et al. (2005) Morwitz et al. (2007) Morwitz (2014)
Definitions and Operational Boundaries of Consumer Motivation
Consumer motivation refers to processes through which an agent assigns priority to an acquisition-related goal and mobilizes cognitive or behavioral resources toward it. It can be operationalized through choice effort, willingness to search, goal persistence, approach behavior, relative expenditure, and direct reports of reasons or desire. None of these indicators alone identifies a single underlying cause. A purchase motivated by hunger, functional necessity, social belonging, aesthetic interest, fear of exclusion, or anticipated savings may produce the same transaction. Conversely, the same reported desire may produce very different search, spending, and substitution patterns depending on available alternatives. Researchers should therefore distinguish the content of a goal from its strength, perceived attainability, and regulatory organization. Deci & Ryan (2000) Bettman et al. (1998) DellaVigna (2009)
Self-determination theory distinguishes different qualities of goal regulation and considers the roles of autonomy, competence, and relatedness. Its broader motivational constructs offer a way to question whether shopping is pursued as enjoyable exploration, for valued personal reasons, or under experienced pressure. These categories require appropriate validated measurement and should not be inferred mechanically from the goods bought. A luxury product can serve multiple motives; a routine household purchase can have social and self-regulatory meanings. The neuroscience of reward likewise separates hedonic impact, cue-triggered incentive value, and learning. Berridge and Robinson's analyses offer a warning against treating a strong desire for a product as proof that its actual consumption will be pleasant. The empirical task is to establish which process a given experiment or behavioral trace identifies. Deci & Ryan (2000) Berridge & Robinson (1998)
Motivational research distinguishes relatively enduring goal priorities from transient states. A planned equipment replacement may remain relevant for months, whereas a display, promotion, hunger state, or social comparison can temporarily increase a product's salience. Even when these states interact, temporal precision matters. A survey fielded shortly after an advertisement may record attention and positive affect rather than a durable commitment to purchase. Observing later payment, cancellation, return, or repeat purchase changes the interpretation of what was initially measured. High motivation may increase search and comparison without increasing immediate buying when consumers defer commitment until a better price arrives. Scientific explanation should model the conversion of interest into specific action rather than equating activity with conversion. Bettman et al. (1998) Dhar (1997) Morwitz (2014)
Purchase Intention, Desire, Expectation, and Behavioral Commitment
Purchase intention is typically measured by asking how likely, willing, or determined a respondent is to buy an identified product or service within a stated interval. Those questions need careful separation. Desire records wanted outcomes; willingness expresses openness to acting under certain conditions; intention refers more directly to intended performance; self-prediction may incorporate obstacles the respondent expects to encounter. A person can strongly desire a product and accurately predict that they will not purchase it because the price exceeds their budget. Someone can expect to buy household essentials without feeling strong desire. Combining all these responses into one undifferentiated score limits construct validity and can mask differences in how particular measures relate to transaction data. Sheeran (2002) Sheeran & Webb (2016) Morwitz (2014)
Behavioral commitment has further dimensions. Intentions may be specific or general, stable or changeable, deliberative or weakly considered, conditional or unconditional. A planned purchase of a named laptop at a named retailer within a week is easier to link to transaction records than a vague favorable attitude toward technology purchases over the next year. Conditional intentions may be explicit, such as buying if a price falls below a threshold, or implicit in a respondent's understanding of the question. Survey design should therefore record the target product, quantity, channel, timing, price assumptions, and availability where these factors determine the outcome. Greater specificity can increase predictive utility while narrowing the interpretation to a particular opportunity. Ajzen (1991) Morwitz et al. (2007)
The language of conversion adds a commercial metric to this psychological framework. A conversion rate is the proportion of an explicitly defined eligible population that completes a specified action during a specified observation period. It is not a psychological trait. A retailer might define conversion as an order placed, a payment settled, a new customer acquired, or a subscription activated. Each definition yields a different dependent variable. The denominator may be unique visitors, sessions, exposed users, eligible households, or surveyed respondents. The scientific connection between intention and conversion requires that the denominator correspond to the population in which intentions were measured and that the event be observable across the relevant channels. Without those conditions, a forecast can appear to fail simply because its operational definitions are mismatched. Morwitz (2014) Chandon et al. (2005)
The Architecture of Consumer Action: Opportunity, Search, Choice, and Exchange
A completed purchase often depends on a sequence of opportunities: encountering a need or cue, recognizing alternatives, acquiring information, evaluating price and product features, reaching a seller, supplying payment and delivery information, and deciding to finalize the exchange. These activities can occur rapidly, overlap, or be delegated, and their order is an empirical question rather than a universal funnel. Research on constructive choice shows that consumers develop preferences while considering alternatives and responding to task demands; their valuation can therefore change during search. A measurement at one stage predicts a later stage with error because the decision maker can encounter new information, revised goals, changed prices, or additional options. Bettman et al. (1998) Dhar (1997)
The purchasing environment defines a set of affordances: actions the person can realistically carry out given the interface, payment method, product availability, delivery region, legal eligibility, and other practical conditions. Two consumers who report identical buying intentions may face different objective opportunities. Even one person may encounter a changed opportunity between intention measurement and checkout. A product becoming unavailable is a failure of opportunity, not necessarily an abandonment of preference. Researchers studying the intention–action relation should measure stock interruptions, access restrictions, shipping cost, time limits, and financing options where possible. Perceived control can affect intention formation, but realized control is established by the environmental conditions and actions that follow. Ajzen (1991) Sheeran & Webb (2016)
Abstention, substitution, delay, and exit are theoretically informative outcomes. In experiments on choice deferral, Dhar showed that uncertain comparative attractiveness can make the no-choice option more appealing. A shopper who leaves without buying may still have invested effort in evaluation and may return when information or alternatives change. Such cases require follow-up rather than categorizing every immediate nonpurchase as failed motivation. Likewise, obtaining a similar product elsewhere is behaviorally distinct from avoiding acquisition altogether. Good research follows the person's acquisition goal across channels and across time, while keeping each event separately identified. The outcome space should include purchase, deferral, substitution, cancellation, and return when scientifically relevant. Dhar (1997) Morwitz et al. (2007)
The Theory of Planned Behavior and Its Explanatory Commitments
The theory of planned behavior organizes deliberative action around attitudes toward performing the behavior, perceived social expectations, and perceived behavioral control. These variables are theorized to contribute to intention, while intention and control are considered predictors of later performance. Applied to consumer acquisition, the target must be behavioral: buying a particular item through an available channel within a meaningful time period. An attitude toward a brand does not necessarily carry the same implications as an evaluation of spending money on that brand today. Perceived normative approval may increase the declared desirability of a purchase yet have little consequence when acquisition is private or when the consumer lacks money. The framework is most informative when its constructs are independently measured and each predicted pathway is tested against transactions rather than a second attitudinal scale. Ajzen (1991) Armitage & Conner (2001)
Armitage and Conner's 2001 meta-analytic review drew on 185 independent studies and found that the planned-behavior variables explained meaningful, incomplete variance in reported intention and behavior. They also reported stronger behavioral prediction with self-reported outcomes than with objective or observed outcomes. This difference is critical for purchase science: a respondent who reports an intention at baseline and later reports having purchased may introduce shared response styles, selective memory, or a desire for consistency. The conceptual prediction remains worthwhile, but validation against receipts, point-of-sale logs, payment confirmations, or independently coded acquisition events establishes a more demanding criterion. Quantitative estimates drawn from broad behavior literatures should not be transplanted as universal coefficients for every retail category. Armitage & Conner (2001) Morwitz (2014)
Criticism of planned-behavior approaches concerns both explanatory completeness and causal inference. Sniehotta, Presseau, and Araújo-Soares argued in 2014 that decades of prediction-oriented research had exposed significant limitations in the framework, prompting substantive replies and extensions. The disagreement illustrates a general research problem: predicting who will act is not equivalent to identifying the intervention that will make a person act. Consumer attitudes, subjective norms, and intentions may share common antecedents; a large association may therefore remain when no single component can be manipulated effectively. Tests should compare planned-behavior predictions with plausible competitors, include objective opportunity and past behavior, and specify what would count as a failed prediction. Sniehotta et al. (2014) Ajzen (1991) Webb & Sheeran (2006)
Constructed Preference, Deliberation, and Consumer Goals
Consumer goals are often refined during the very process of shopping. Bettman, Luce, and Payne proposed a constructive account in which decision strategies depend on task complexity, available information, processing constraints, and the decision maker's objectives. A consumer initially considering a practical device may discover trade-offs involving battery life, repairability, price, and appearance and then revise the criterion that matters most. Purchase intention can accordingly change without any psychological inconsistency. What looked like a firm intention at the beginning of search may have been a provisional ranking based on incomplete attributes. Studies that measure intention only once cannot distinguish a failure of execution from legitimate preference revision after new evidence. Bettman et al. (1998) Morwitz (2014)
Choice deferral provides a revealing test of this account. Dhar's experiments investigated circumstances in which the available alternatives were sufficiently close in attractiveness that the consumer preferred to wait rather than commit. The no-choice option is a genuine behavioral response, not missing data. If a researcher forces a choice in a laboratory but observes an unconstrained market later, measured preferences can appear more decisive than natural buying decisions. Including an explicit deferral alternative, recording confidence, and separating information-search motives from readiness to pay improve ecological validity. The analysis also distinguishes costly indecision from economically sensible waiting, since new information or a future discount may change the expected value of acquisition. Dhar (1997) Bettman et al. (1998)
Motivation can be directed toward goals other than immediate acquisition. A shopper may seek reassurance, mastery of a product category, entertainment, social identity, or price knowledge. These goals can generate the same preliminary behaviors as purchase-directed motivation: browsing, comparisons, saving favorites, or filling a cart. Researchers should elicit the consumer's proximal objective and examine whether the measured activity fulfills that objective. The usefulness of the funnel metaphor depends on allowing different pathways and reversals; digital navigation data alone rarely reveal the meaning of a click. Choice construction, desire, and transaction commitment should be modeled as distinct potential processes with separate observables. Close & Kukar-Kinney (2010) Kukar-Kinney & Close (2010) Bettman et al. (1998)
Learning, Incentive Salience, and Repeated Purchasing
Repeated buying may arise through different processes. A consumer can repeatedly choose a product after evaluating its price and benefits, or repeated encounters with a familiar context can trigger a response with little fresh deliberation. Ouellette and Wood's synthesis distinguished well-practiced actions in stable contexts from actions that require renewed intentional control. Wood and Neal's subsequent account described habits as context-response associations that interface with goals. These theories help explain why the same item may be bought under strong deliberative intention during the first trial and with relatively weak conscious planning on later occasions. Observing repetition alone does not identify habit strength; researchers need evidence concerning cue dependence, automaticity, and the stability of contexts. Ouellette & Wood (1998) Wood & Neal (2007)
Incentive-motivation research further separates wanting, liking, and learning. Berridge and Robinson asked whether dopamine-related reward mechanisms are better understood through pleasure, associative prediction, or incentive salience, emphasizing that these functions are distinguishable. Applied cautiously to consumer research, this distinction discourages the claim that a product prominently sought in a digital interface must be equally satisfying after purchase. Marketing cues can make an object salient, and consumption experience can subsequently alter evaluation; neither implies a simple one-to-one neurological explanation for a shopping session. Human purchase decisions require additional measurement of beliefs, resources, social meaning, and actual exchange. Neurobiological models are useful when they generate appropriately scoped predictions rather than decorative explanations. Berridge & Robinson (1998) Morwitz (2014)
Repeat purchases also complicate causal attribution in marketing datasets. Past purchasing predicts future acquisition partly because of preferences and needs, partly because of stable access and constraints, and sometimes because of habitual cue-response patterns. If a model treats every repeat transaction as the effect of a newly measured intention, it exaggerates the role of deliberation. If it assigns all repeat buying to habit, it overlooks periodic reassessment and price sensitivity. Designs can use context changes, discontinuities in assortment, and within-person observation to examine whether action persists under changed circumstances. Distinguishing goal-directed evaluation from automatic repetition is a shared concern with the neighboring canonical dossier on goal-directed action and habit formation. Wood & Neal (2007) Ouellette & Wood (1998)
Time, Delayed Gratification, and Changes in Purchase Readiness
The time between an intention report and an opportunity to buy changes what prediction means. A short interval offers fewer occasions for preferences, product availability, or household resources to change. Longer horizons expose respondents to competing purchases, unexpected expenses, promotions, social recommendations, and shifting priorities. Morwitz, Steckel, and Gupta found stronger intention-purchase correspondence for shorter rather than longer time horizons across varied settings. This is a conditional empirical finding, not a universal prediction that intention inevitably decays. A person may form an advance plan precisely because a future purchase is important. Researchers should record the interval and compare predictions at matched temporal distances, distinguishing the timing of the target behavior from the timing of measurement. Morwitz et al. (2007) Morwitz (2014)
Models of present-biased and dynamically inconsistent choice provide one explanation for departures between long-term plans and immediate decisions. Laibson's 1997 model of hyperbolic discounting analyzes how relative valuation can shift as an option becomes immediate and how commitment devices can constrain future choices. This framework is relevant to acquisition decisions when a consumer plans to save but later experiences a tempting purchase opportunity, or intends to buy a durable product but reallocates resources to immediate needs. It is a formal model with testable behavioral implications; a particular consumer's failure to follow a plan does not establish hyperbolic discounting without competing explanations and repeated choice observations. Laibson (1997) DellaVigna (2009)
Commitment can sometimes be made operational through plans that specify when and how action will occur. Gollwitzer and Sheeran's meta-analysis of implementation intentions addresses whether if-then plans improve attainment of goals under self-regulatory difficulties. The broader intervention evidence indicates that linking a situational cue to an intended response can facilitate action in many settings. Translating this into a purchasing context requires safeguards: an effective reminder or one-click completion may increase transactions, but the welfare consequences depend on whether the purchase serves a consumer-endorsed goal. Research should distinguish improved implementation of a settled intention from externally induced urgency or impulsive buying. The measured endpoint remains completed behavior, accompanied when appropriate by satisfaction, regret, returns, and spending consequences. Gollwitzer & Sheeran (2006) Sheeran & Webb (2016)
Measuring Purchase Intentions: Questions, Scales, and Temporal Reference
An intention measure is a forecast only when its response format specifies the action whose occurrence can later be checked. Asking whether a respondent likes a product, might consider buying it, intends to purchase it, or expects to have purchased it by a specified date produces different information. Response scales may be verbal likelihood categories, numerical probabilities, commitment ratings, or stated plans. Each instrument embeds assumptions about uncertainty and respondents' ability to predict future resources and preferences. A scale is useful when its interpretation is sufficiently stable across respondents and repeated observations to support external validation. The temporal reference, price assumption, product specification, and purchase channel should be recorded alongside the response. Studies should report the precise wording rather than treating the label 'purchase intention' as a complete operational definition. Morwitz (2014) Sheeran (2002)
Predictive validity requires comparison against an independently defined acquisition endpoint. If the target is buying a particular model within thirty days, subsequent substitution with a cheaper model should be coded separately from buying the original target. A respondent who postpones acquisition until day thirty-five falls outside the defined event window even when the underlying goal persists. Analysts can examine both a prespecified binary endpoint and time to acquisition, preserving the distinction between a delayed purchase and indefinite nonpurchase. Attrition becomes especially consequential when people who decline follow-up are also less likely to act. Objective administrative records, where available and lawfully linkable, can reduce outcome misclassification but require checks on whether off-platform purchases remain invisible. Morwitz et al. (2007) Morwitz (2014)
Reliability and validity belong to different layers of the measurement problem. A respondent can repeatedly give an identical intention answer that predicts a transaction poorly because unexpected prices and constraints change. Conversely, unstable intention reports can be accurate descriptions of rapidly changing goals. Internal consistency of several similar questionnaire items does not establish behavioral validity, and an apparently strong correlation with another attitude measure may reflect shared survey response tendencies. For forecasting, researchers should evaluate calibration, discrimination, and time-specific error against actual behavior. For psychological explanation, they must also distinguish the influences that change an intention from the influences that help implement it. Chandon et al. (2005) Webb & Sheeran (2006) Morwitz (2014)
Observing Purchases: Transactions, Panels, and Follow-Up Data
Observed purchasing encompasses different record types, each with a distinct measurement domain. A point-of-sale receipt can verify a transaction at one seller, a loyalty account can connect repeated purchases made with that identifier, a payment record can indicate expenditure without necessarily identifying the product, and a participant diary can provide information across stores while reintroducing reporting error. E-commerce logs can capture browsing, cart modification, checkout attempts, successful payment, returns, and cancellations. Researchers must establish which event constitutes acquisition. A completed checkout may later be refunded; a payment authorization may fail; a gift acquisition may fulfill a household rather than an individual purchase goal. The choice of endpoint can change both the estimated intention–behavior association and the interpretation of consumers' actions. Morwitz (2014) Kukar-Kinney & Close (2010)
Linking baseline intentions to later transaction records makes an important advance over cross-sectional survey comparisons, but identification remains conditional on participation and data coverage. A retailer may see only purchases at its own store, while a survey respondent can acquire the product through another channel. If linkage requires loyalty-card use or consent to tracking, the linked sample may differ systematically from the population of shoppers. Product matching, household sharing of accounts, ambiguous returns, and observation-window boundaries all require predefined rules. Strong designs report the number of eligible survey respondents, the number successfully linked, the extent of missing observations, and how conclusions change under alternative defensible coding decisions. Morwitz et al. (2007) Hainmueller et al. (2015)
Panel measurements permit analysis of intention stability, changes in resources, and within-person shifts, but repeated questioning can influence the phenomenon. Morwitz, Johnson, and Schmittlein demonstrated that measuring intentions can affect subsequent behavior. This means an apparently passive forecasting design may be an intervention, especially when purchase questions increase attention to an option or prompt respondents to formulate a plan. A comparison group that receives no purchase-intention question can help estimate this reactivity. When the act of measurement changes shopping, predictive performance calculated only among questioned respondents should not be generalized automatically to a population that has never been surveyed. Morwitz et al. (1993) Chandon et al. (2005)
The Intention–Behavior Gap: Description, Mechanisms, and Explanation
The intention–behavior gap is the divergence between what individuals report intending to do and what they subsequently do within the relevant observation period. It can be described in several ways: proportions of strong intenders who do not act, differences between planned and realized purchase frequency, forecast-calibration error, or incomplete association between intention ratings and observed transactions. These measures should not be confused. A correlation can be high even when predicted aggregate sales are too optimistic, and a well-calibrated probability forecast may still leave many high-intention individuals without a purchase. The scientific question is which inferential problem is under study: individual classification, aggregate forecasting, or explanation of nonenactment. Sheeran (2002) Morwitz (2014)
Psychological and material mechanisms can generate the same observed gap. Delays and competing goals, failure to encounter a suitable opportunity, changes in preferences during search, insufficient resources, and unanticipated transaction friction all prevent a previously plausible plan from reaching completion. Socially desirable reporting can instead generate an intention answer that never represented the respondent's likely action. These possibilities demand different tests. A design that records only an intention survey and later purchase cannot establish which mechanism was responsible; additional observations on opportunity, financial constraints, switching, actual price, and deliberation are needed. Conceptually, the gap is an explanandum requiring decomposition, not an autonomous psychological force that causes nonpurchase. Sheeran & Webb (2016) Ajzen (1991) Bettman et al. (1998)
The gap also has an asymmetric structure. A respondent may buy without previously reporting a strong intention because of a sudden need, an attractive immediate offer, exposure to a new item, a social obligation, or an established routine. Analyses focusing only on disappointed intenders omit this other source of forecast error. A useful contingency table separates intended purchases that occurred, intended purchases that did not occur, initially unintended purchases that occurred, and initial nonintentions followed by nonpurchase. Researchers can further distinguish whether an unexpected transaction represents a preference change, cue-triggered action, or a practical response to a newly arising need. Such classification prevents an intention-centered model from defining every consumer action in terms of the same single antecedent. Ouellette & Wood (1998) Morwitz (2014)
Intention Change and Action Change: Evidence from Experimental Synthesis
Interventions that increase stated intentions cannot automatically be assumed to increase subsequent purchasing. Webb and Sheeran's 2006 meta-analysis of experimental studies compared the effects of experimental manipulations on intentions with their effects on behavior, finding that appreciable changes in intention translated into smaller changes in observed action. The result belongs to a broad behavioral literature rather than a retail-only treatment estimate. Its methodological importance for consumer psychology is the distinction between successful persuasion and successful enactment. A campaign might improve attitudes and reported willingness to buy while leaving real-world sales constrained by price, logistics, or substitute options. Intervention evaluation therefore requires separately designated intention and purchase outcomes. Webb & Sheeran (2006) Sheeran & Webb (2016)
Even when a manipulation affects both constructs, the mediation claim needs care. An advertisement may increase intention and also make a retailer easier to find; a discount may increase intended purchasing and simultaneously lower the cost of purchase. An observed association between these post-treatment variables does not prove that the change in intention transmitted the entire effect. To support causal mediation, investigators need a credible identification strategy, temporal ordering, and assumptions about common causes of the mediator and outcome. Randomizing the intervention identifies an overall effect under appropriate conditions, but it does not by itself randomize intention. Distinguishing intervention efficacy from mechanism attribution is especially important when promotion, product placement, or checkout simplification changes several pathways together. Webb & Sheeran (2006) DellaVigna (2009)
Implementation-oriented interventions target the process between a goal and its execution. If-then planning, a carefully timed reminder, a saved budget, a shopping list, or reduced administrative friction might make an already endorsed acquisition more feasible. Yet a rise in transaction frequency is not a sufficient evaluation of consumer benefit. In categories involving debt, unnecessary recurring charges, or impulsive buying, the same increase may reflect reduced deliberation rather than improved goal attainment. Research should assess the consumer's prior goal, informed approval of the intervention, and downstream outcomes such as regret, returns, or financial strain where appropriate. Distinguishing motivational change from execution support makes the mechanisms more precise and the welfare consequences more interpretable. Gollwitzer & Sheeran (2006) Sheeran & Webb (2016)
The Survey as Intervention: Question–Behavior and Self-Generated Validity Effects
Questioning can alter which options become salient and which commitments a person rehearses. Morwitz and colleagues' early work on whether measuring intent changes behavior established that purchase-intention surveys may influence later market action. A respondent who has just considered buying a durable good can subsequently notice related advertising or become more willing to search. This measurement effect should be separated from the predictive accuracy of the respondent's answer. If only people who were asked the question are observed, the survey may be forecasting an intervention-created population state. The design solution is to compare otherwise similar groups differing in whether and how the question was asked while monitoring purchases through an independently available outcome system. Morwitz et al. (1993) Chandon et al. (2005)
Chandon, Morwitz, and Reinartz examined the extent to which survey elicitation itself strengthens apparent intention–behavior consistency. Their analysis of self-generated validity connects measurement reactivity to a broader psychometric problem: answering can make the evaluated concept more cognitively available and therefore more likely to guide later action. The consequences depend on the studied behavior and measurement setting, so the existence of a question–behavior effect does not imply that every intention survey meaningfully increases sales. It means researchers should treat the survey procedure, the timing of exposure, and the presence of nonquestioned comparison observations as part of the causal design. Calibration performed under one measurement regime may not transfer to another. Chandon et al. (2005) Morwitz (2014)
Ethically, measurement reactivity raises an issue for research partnerships with retailers. A company may want a commercially useful forecast, while investigators aim to measure naturally occurring behavior. If the survey encourages purchases, the instrument may simultaneously serve commercial interests and alter the consumer's options. Transparent study documentation should explain whether data collection was observational or promotional, how questions were worded, and whether assignment to measurement was randomized. When results are used to evaluate advertising efficacy, researchers should report separately the consequences of the advertisement and the consequences of asking about its products. This is a concrete case in which the scientific measurement process becomes part of the market context under study. Morwitz et al. (1993) Chandon et al. (2005)
Prices, Search Costs, Salience, and Realized Demand
A consumer's reported desire can coexist with a price threshold that prevents acquisition. Price changes do more than alter an abstract evaluation; they determine whether purchase fits available income, competing obligations, or the alternatives on offer. Studies of price salience illustrate the difference between noticing and paying a cost. Chetty, Looney, and Kroft investigated whether consumers respond differently to taxes incorporated into posted shelf prices and taxes added at the register. Their field evidence on tax-inclusive displays demonstrates that visibility of the effective price can change observed demand. This is a behavioral response measured in a market setting, rather than an inference from hypothetical willingness to pay. The mechanism under discussion includes attention to full cost, not simply an underlying preference for the product. Chetty et al. (2009) DellaVigna (2009)
Pricing interventions are informative because they can influence opportunity, motivation, and transaction at once. A discount may make a previously unaffordable purchase feasible, increase attention to a product, or shift a comparison against competitors. Interpreting the effect solely as increased desire would overlook these additional channels. A scientifically strong design identifies the treatment, measures the resulting price actually faced, and observes whether purchases are displaced from alternative products or future periods. Changes in total spending, transaction count, and purchases of a single promoted item need not coincide. For studies linking stated willingness to buy with actual demand, the question should specify whether the forecast assumes a particular price, discount, tax, shipping fee, and return policy. Chetty et al. (2009) Morwitz et al. (2007)
Search costs can also create an apparent intention–behavior gap. A person can be willing to buy at a given price yet abandon the process when product comparisons, registration requirements, or unclear fees consume substantial time and attention. Conversely, easy one-click transactions can enable purchases that were weakly planned. Search and interface data help identify the pathway, but their interpretation requires caution. The number of product pages viewed may reflect active comparison, leisure browsing, uncertainty, or difficulty finding adequate information. A causal account must examine experimentally varied task conditions or observed natural changes in search costs while tracking the consumer's eventual acquisition outcome. Bettman et al. (1998) Close & Kukar-Kinney (2010) Dhar (1997)
Product Categories, Purchase Frequency, and Forecast Horizons
Purchase intentions have different forecasting properties across products, consumers, and decision contexts. Infrequently purchased durables may involve lengthy planning, dependence on household finance, and opportunities to delay or substitute. Routine consumables may be purchased through established habits or replenishment needs and can be less dependent on explicit plans. For novel products, intentions often reflect limited information and expected future availability; for established brands, prior experience can supply more stable expectations. Morwitz, Steckel, and Gupta compared conditions influencing intention–sales prediction and reported systematic differences associated with purchase interval and product characteristics. Rather than transferring a single correction factor between categories, analysts should test predictive performance within sufficiently comparable markets. Morwitz et al. (2007) Morwitz (2014)
The survey's time horizon and the consumer's natural purchase cycle must be aligned. Asking a shopper whether they intend to buy a refrigerator next week has a different base rate than asking whether they expect to buy one over several years. A high fraction of people who correctly predict nonpurchase can make a model appear accurate even when it poorly identifies rare purchases. Researchers should report class prevalence, calibration within meaningful intention groups, and performance against a baseline using prior purchase, need, and available opportunity. When sales follow seasonal patterns, time, region, and market changes may explain some apparent predictive success. This is why out-of-sample testing should respect calendar order and avoid mixing the same consumer's future record into the model's training period. Morwitz (2014) Ouellette & Wood (1998)
Experience can improve a consumer's ability to forecast familiar needs, but expertise can also change what is understood by the survey question. A buyer familiar with the market may read an intention item as a conditional plan tied to price, while a novice may interpret it as a statement of desire. The two answers may look equivalent on a numeric scale while encoding different assumptions. Studies across categories and populations should conduct cognitive interviews, inspect distributional differences, and test whether the intention measure relates similarly to subsequent purchases in each group. Measurement comparability is a prerequisite for interpreting group-level forecast differences as psychological differences. Sheeran (2002) Morwitz et al. (2007)
The Digital Shopping Cart and the Ambiguity of Purchase Readiness
The online shopping cart is frequently treated as a visible indicator of impending purchase, but its functions extend beyond checkout. Close and Kukar-Kinney studied motivations behind cart use and argued that carts serve exploratory and organizational roles in addition to acquisition. Consumers may use a cart as a temporary comparison board, wish list, price-monitoring tool, or way to compute shipping charges. Treating every addition as an intention to transact produces false-positive purchase forecasts. A cart is an interface state generated by user action; its psychological meaning depends on what the user intends to achieve and how the site structures the shopping process. Behavioral research should distinguish cart building, repeated revisits, checkout initiation, and finalized payment. Close & Kukar-Kinney (2010) Kukar-Kinney & Close (2010)
The distinction becomes more important when the same consumer can browse across devices, sessions, and merchants. An abandoned cart at one retailer can coincide with a successful purchase elsewhere or a deliberate decision to wait for a better offer. Cookie loss, login differences, household devices, and privacy restrictions can prevent accurate person-level linkage. In a study using e-tail clickstream data, Kukar-Kinney and colleagues examined shopping-cart abandonment with behavioral traces, providing a framework for examining the digital process rather than relying solely on retrospective intention ratings. Clickstream models still require validation against purchase events and careful treatment of events outside the observed seller's system. The research endpoint should define whether the object is a cart, a session, a buyer, or an acquisition goal. Kukar-Kinney et al. (2022) Morwitz (2014)
Cart abandonment is also a poor direct measure of consumer dissatisfaction. A person may leave because a shipping estimate was unexpectedly high, payment verification failed, product availability changed, or the cart had always served as a planning aid. These explanations lead to different interventions and welfare assessments. Reducing checkout friction might help consumers carry out settled choices; pressuring visitors with countdown messages might increase sales while reducing decision quality. Experimental tests can compare changes in checkout completion with subsequent returns, complaints, and repeat satisfaction. Such outcomes help separate a purely commercial optimization target from a psychological explanation of intended and realized consumer behavior. Kukar-Kinney et al. (2022) Close & Kukar-Kinney (2010)
Digital Pathways, Multi-Channel Shopping, and Purchase Sequences
Retail environments generate observable sequences that can be modeled as transitions among exposure, information search, cart creation, checkout, payment, and postpurchase activity. These transitions are useful descriptions of what happened, but a sequence model alone does not show why each action occurred. An advertisement impression may precede purchase because a user already had a purchasing goal and encountered targeted marketing. A recommendation click can indicate curiosity rather than the formation of a new commitment. To study motivation, researchers must pair behavioral traces with interventions, carefully timed self-reports, or other measures that identify goals and beliefs. The methodological advantage of fine-grained digital records lies in temporal resolution, provided that identities, event definitions, and observation boundaries are accurately established. DellaVigna (2009) Morwitz (2014)
Consumers now move among marketplaces, physical stores, direct-to-consumer websites, and social platforms. A single seller's data reveal a fraction of that ecosystem. A study of conversion following a marketing email might miss purchases in-store or at a competitor, and a customer panel might omit visitors who decline data collection. Reconstructing an acquisition trajectory requires explicit linkage rules and sensitivity analysis for missing channels. Where broad linkage is impossible, scientific conclusions should be restricted to observed transactions within the particular ecosystem. Strong experimental designs randomize exposure before the action of interest and report the eligible population, assignment process, attrition, and outcome-collection window. Morwitz et al. (2007) Chandon et al. (2005)
Temporal analysis can reveal that the relation between intention and purchase varies with elapsed time. A person might buy quickly after a new household need, postpone a planned discretionary purchase, or return repeatedly to evaluate product alternatives. Longitudinal models can estimate time to purchase and competing events such as substitution or cancellation. Their interpretability depends on the censoring assumption: if the observation system stops following someone who quietly purchases elsewhere, treating that record as ordinary nonpurchase can bias estimates. The relevant question is whether the model predicts an observed event in its defined population, not whether the complete psychological process can be reconstructed from a platform log. Morwitz (2014) Kukar-Kinney et al. (2022)
Ethical Labels, Values, and Purchases Observed in the Field
Values-based consumption offers an especially clear example of the difference between a favorable stated attitude and actual expenditure. Consumers may report supporting fair labor standards, environmental protection, or responsible supply chains while confronting higher prices, uncertain claims, and competing household needs during shopping. These pressures can cause a discrepancy without establishing hypocrisy or a lack of sincerity. Carrington, Neville, and Whitwell proposed a conceptual framework for explaining the ethical purchase intention–behavior gap and highlighted planning, control, and real-world context. Its importance is explanatory organization; specific causal claims still require experimental or observational tests against verified purchases. The specialized science of sustainability-related discrepancies has its own canonical research object, while the present dossier considers the general purchase pathway. Carrington et al. (2010) Sheeran & Webb (2016)
Hainmueller, Hiscox, and Sequeira conducted a multistore field experiment examining demand for fair-trade coffee. Their observed sales outcomes, including responses to labeling and price conditions, demonstrate how market interventions can be evaluated beyond hypothetical ethical preferences. Field experiments of this kind increase behavioral relevance by allowing shoppers to face real alternatives and real expenditure, although causal interpretation applies to the tested treatment, retail settings, and observed customer composition. Whether an observed sales response reflects self-image, product-quality inference, moral commitments, or another factor requires measurements or tests that differentiate those pathways. A single ethical label may communicate several attributes simultaneously. Hainmueller et al. (2015) Bettman et al. (1998)
The literature on ethically motivated consumption also warns against treating high-intention nonbuyers as a homogeneous group. Consumers vary in knowledge of standards, trust in certifications, perceived efficacy of purchases, price constraints, and how they interpret responsibility. Some may purchase a substitute product supporting their values through an unobserved channel, while others may explicitly prioritize different considerations at the point of sale. Studies should examine the compatibility of intended action, observed alternative, available price, and the consumer's own goal. What matters for the general psychology of purchasing is the interaction between motive, feasible choice set, and final exchange. Moral assessment of the buyer is an external normative question that should not be substituted for the empirical account. Carrington et al. (2010) Hainmueller et al. (2015)
Implementation Supports: Planning, Reminders, and Feasible Action
A consumer who has decided to acquire a necessary item may still fail to act because the relevant moment passes, the option is difficult to access, or a competing task receives attention. Implementation-intention research studies whether specifying a cue and a response helps people enact goals they already possess. Gollwitzer and Sheeran's meta-analysis synthesizes such interventions across behavioral domains, emphasizing the link between anticipated situations and planned actions. For consumer research, the theoretically relevant question is whether an advance commitment such as 'when I arrive at the store, I will compare the labeled energy ratings' changes what is bought, rather than merely improving memory for a shopping intention. Intervention success must be shown against observable events and interpreted within the tested product, setting, and population. Gollwitzer & Sheeran (2006) Sheeran & Webb (2016)
Randomized reminder and planning experiments outside retail supply useful design analogies but do not provide automatic estimates for consumer buying. Milkman and colleagues, for example, investigated planning prompts in an influenza-vaccination context with an observed behavioral outcome. Their work illustrates the value of identifying a concrete opportunity, randomizing prompt exposure, and following whether an action occurs. A purchase intervention could similarly compare reminders, saved lists, or explicit collection plans while recording actual acquisition and alternative outcomes. The external validity question is central: obtaining a vaccine, renewing essential supplies, and purchasing discretionary merchandise involve different incentives and welfare considerations. Evidence for a planning mechanism should therefore be tested in the purchasing setting of interest. Milkman et al. (2011) Gollwitzer & Sheeran (2006)
Opportunity interventions can remove practical barriers without changing a person's underlying desire. Making accurate delivery dates visible, offering accessible payment methods, providing a clearly priced pickup option, or preserving a saved cart may make an intended purchase feasible. These interventions also affect trust and perceived price, so designs should measure intermediate experiences rather than assume one pathway. A seller might observe higher completion but also higher return or cancellation rates, indicating that the initial endpoint was too narrow. Consumer-oriented evaluation should ask whether actions matched established preferences, whether people understood the transaction, and whether the intervention imposed costs on those who did not purchase. The most informative results encompass both immediate behavior and relevant postpurchase outcomes. Ajzen (1991) Kukar-Kinney et al. (2022)
Choice Architecture, Defaults, and the Evaluation of Consumer Outcomes
Choice architecture describes the organization of available options, descriptions, prompts, and default settings. A preselected subscription term, a prominently displayed delivery option, or a sequence that makes cancellation difficult can influence observed behavior without substantially changing a measured intention. Defaults are therefore an important test case for the distinction between an internally endorsed action and a transaction produced by environmental structure. Experimental research on defaults in other contexts, such as Johnson and Goldstein's organ-donation comparison, established their relevance to behavior but cannot be generalized mechanically to financial transactions or consumer consent. Retail experiments must identify the choice presented, the consequences of inaction, and the consumer's opportunity to understand and reverse the selection. Johnson & Goldstein (2003) DellaVigna (2009)
An intervention that raises the number of completed purchases may be commercially successful while leaving a consumer's goals less well served. Automatic renewal, hidden fees, time pressure, or confusing opt-out conditions can increase transaction metrics while reducing the quality of choice. A welfare evaluation therefore requires an explicit objective and indicators beyond conversion. Possible indicators include whether the consumer had an independently reported prior goal, knew the terms, retained a realistic ability to refuse, experienced regret, initiated a return, or sustained an unwanted charge. No single postpurchase measure settles welfare, but several complementary observations can prevent an analysis from equating sales growth with psychological benefit. Ajzen (1991) DellaVigna (2009)
Choice architecture also creates methodological problems for studies that compare expressed intention with observed purchasing. Suppose two interfaces produce identical baseline intention scores but different checkout rates. The observed gap could be caused by a default, not by a difference in motivational strength. If respondents see the interface before answering, the design might also shift the intention measure itself. Random assignment, clear temporal ordering, and separate measurement of intention and behavior can identify which differences arise from the environment. Behavioral records permit stronger claims when researchers describe consent, treatment exposure, off-platform purchase possibilities, and whether an intervention changes the definition of the outcome. Chandon et al. (2005) Webb & Sheeran (2006)
Experimental Design, Field Identification, and Mechanism Testing
Experimental work can establish whether a particular intervention changes observed acquisition when assignment produces comparable groups and the outcome is measured consistently. A factorial design might independently vary full-price disclosure and ease of checkout; the resulting comparisons estimate their separate and joint effects on purchasing within that study. However, a causal effect of price display on sales does not by itself identify increased motivation, decreased cognitive load, or a change in perceived fairness. Testing these mechanisms requires additional measures and, where feasible, interventions that distinguish their predictions. The strongest consumer experiments state the unit of assignment, define the intended population, avoid post-treatment selection, and report effects on the predesignated outcome. Chetty et al. (2009) DellaVigna (2009)
Natural experiments and quasi-experimental studies exploit market changes such as a tax adjustment, store opening, delivery-area expansion, or administrative policy change. Credible inference depends on whether the change supplies variation plausibly independent of unmeasured determinants of purchasing, or whether assumptions such as parallel trends are supported. Such designs can complement controlled laboratory decisions by studying consequential expenditures under ordinary conditions. They also face product substitution, changing customer composition, anticipation, and the interaction of multiple policies. A statistically significant before–after difference in sales without a defensible counterfactual is evidence of temporal change, not sufficient proof of an intervention's causal influence. DellaVigna (2009) Chetty et al. (2009)
Process measures can clarify how consumers move from intention to acquisition. Researchers can record when product information was viewed, whether the effective price was understood, what options were compared, and at which stage an opportunity disappeared. Yet mediation estimates derived from these processes remain vulnerable to confounding: consumers with greater commitment may both inspect more product details and purchase more often. Experimental manipulation of information access, matched longitudinal measures, and appropriately specified causal models make mechanism claims more credible. The same caution applies to claims that an artificial recommendation 'caused trust,' that a reminder 'created desire,' or that a discount 'removed self-control failure.' Scientific attribution must follow the design's identification strength. Bettman et al. (1998) Webb & Sheeran (2006)
Forecasting, Calibration, and Out-of-Sample Prediction
Predictive models can estimate the probability of a purchase from survey answers, past buying, prices, demographic variables, search patterns, and current availability. The accuracy of such a model is an empirical property of a defined deployment population and observation window. A model trained on active loyalty customers, for example, may perform poorly among first-time visitors or shoppers whose acquisition occurs mainly in competitor channels. Evaluation should separate individuals or households between training and test data where appropriate and preserve the temporal ordering of predictors and outcomes. A future transaction, a return record, or an intention survey conducted after checkout must never leak into the pre-purchase feature set. Morwitz (2014) Morwitz et al. (2007)
Calibration concerns whether stated or estimated probabilities correspond to observed purchase frequencies within relevant groups. Discrimination concerns whether a model ranks those who buy ahead of those who do not; these are different properties. For low-frequency purchases, a model may have high overall accuracy by predicting almost universal nonpurchase while providing little useful information about buyers. Appropriate evaluation includes event prevalence, calibration curves, precision and recall at justified thresholds, proper probability scoring, and comparison with simple baselines based on past purchase or exposure. Where the purpose is economic forecasting rather than individual targeting, aggregate volume error and uncertainty intervals also matter. Choosing a metric after viewing the results creates avoidable optimism. Morwitz (2014) Morwitz et al. (2007)
Prediction is not explanation. A strong purchase classifier using current price, loyalty history, and checkout-stage behavior may outperform a questionnaire-based model without revealing which interventions can help consumers achieve their goals. A model that predicts near-completed checkouts is answering a different question from one that predicts next year's acquisition at the first exposure to a product. Studies should articulate the prediction time, decision horizon, eligible population, observed purchase endpoint, and intended practical use. They should test generalization to future periods, changed prices, new product categories, and groups with different access patterns. Longitudinal stability and sensitivity to policy interventions are necessary before a forecast is used to support claims about enduring consumer psychology. Morwitz (2014) DellaVigna (2009)
Selection Bias, Missing Behavior, and Replication in Consumer Research
Consumer samples are often selected by the very processes under study. A survey of site visitors excludes people who never encountered or reached the site; a loyalty panel excludes shoppers who do not register; a purchase-follow-up sample may exclude those unwilling to share spending information. Conditioning analyses on cart creation or checkout initiation can further select on motivation, interface access, and price sensitivity. Estimates of intention–purchase correspondence can consequently differ across the selected stages without representing changes in the underlying population relation. Scientific reports should map the selection process, state the target population, and describe which generalizations follow from the available evidence. Chandon et al. (2005) Kukar-Kinney et al. (2022)
Missing purchase data require special care. A person with no recorded transaction could have abstained, bought through another channel, paid with an unlinked account, or become unobservable before the study ended. These cases have different implications. When the study is restricted to one seller, analysts should describe the endpoint precisely as a purchase from that seller rather than any purchase of the product category. Sensitivity analysis can examine plausible off-platform behavior, account for attrition where assumptions permit, and distinguish known nonpurchase from unknown outcome status. Technical richness does not compensate for an incomplete observational boundary. Morwitz et al. (2007) Kukar-Kinney et al. (2022)
Replicability in consumer science concerns more than obtaining a similar p-value. Experiments should record products, prices, cultural context, timing, store format, recruitment procedures, treatment exposure, and the economic conditions under which a result was obtained. Field outcomes may legitimately change across recessions, new payment technologies, or altered retail interfaces. Independent replication tests whether a proposed mechanism continues to operate under documented conditions; conceptual replication can investigate its boundary conditions with a different product or population. Publishing null outcomes, uncertainty estimates, and predeclared analysis choices would improve the ability to distinguish robust effects from favorable but selective demonstrations. The evidential quality of each conclusion depends on measurement and design rather than on the prominence of the theory invoked. DellaVigna (2009) Webb & Sheeran (2006)
Resources, Culture, Experience, and Heterogeneity of Consumer Action
Purchase implementation is conditioned by economic resources. Household income, access to credit, transportation, stable housing, available time, and the reliability of delivery infrastructure determine which intentions can be acted on. These conditions also influence what a respondent means when expressing willingness or probability of buying. A consumer with tight financial constraints may intend to purchase only if an expense remains affordable after essentials, while a wealthier consumer can act on the same preference with less budget adjustment. Treating resource-dependent nonpurchase as a uniform failure of willpower substitutes a psychological label for an environmental constraint. Research on consumer intention should measure feasible choices as well as evaluations, and should avoid extrapolating effects from convenient samples to materially different populations. Ajzen (1991) DellaVigna (2009)
Age, experience, household roles, and cultural institutions can change the structure of purchasing. A young adult making an unfamiliar financial commitment encounters different information and responsibilities from a household routinely replenishing familiar goods. Cultural norms may affect the importance of family approval, gift obligations, bargaining, and trust in sellers. Yet categories such as age or country are coarse proxies for mechanisms. Strong comparative research measures the actual norms, decision authority, product access, and purchasing context rather than assigning a psychological cause to demographic membership. Equivalence of intention questions should be established before comparing coefficients across groups, especially when translated terms such as 'likely,' 'plan,' and 'expect' carry different conversational implications. Fishbein & Ajzen (1975) Sheeran (2002)
Consumer heterogeneity can also produce misleading aggregate correlations. A high-intention subgroup might purchase frequently because of both stronger preferences and more reliable access, while a resource-constrained subgroup might report similarly strong intentions but rarely complete a transaction. Pooling the groups can yield an apparent single intention–purchase relation that conceals two different processes. Longitudinal analyses should investigate within-person change as well as between-person differences, and model price, opportunity, and observation coverage at the level at which they operate. Testing interactions is scientifically useful when grounded in a prior mechanism and evaluated with adequate precision rather than through indiscriminate subgroup fishing. Morwitz (2014) Morwitz et al. (2007)
Human–Artificial Decision Assistance and Delegated Consumer Choice
Digital marketplaces increasingly offer algorithmic rankings, recommender systems, conversational assistants, and automated reorder functions. These arrangements alter which information a consumer encounters and, sometimes, which parts of a transaction are delegated. A recommendation system can change the opportunity set by selecting options, framing comparisons, or prepopulating a cart. A consumer may endorse a goal such as obtaining a suitable device and then delegate part of the product search or purchase process to software. The empirical question becomes how the human's goals are represented, whether the recommendation system accurately captures relevant constraints, and how the eventual action relates to the person's authorization. This is a hypothesis-generating extension of established research on constructive choice and environmental affordances; it does not imply that artificial systems possess human desires or intentions. Bettman et al. (1998) Ajzen (1991)
Artificial assistance creates a new measurement problem: who or what generated the recorded action? An automated order executed under a user-defined rule is a completed transaction, but it need not indicate that the person consciously formed a fresh purchase intention at the moment of payment. A conversational recommendation that increases product consideration is also different from an autonomous reorder triggered by a preset inventory threshold. Future research should distinguish human-originated intention, system-generated recommendation, explicit human approval, standing delegated authority, transaction execution, and subsequent evaluation. These variables make it possible to study assistance, automation errors, reversibility, and responsibility without assuming that the same mechanism explains every machine-mediated purchase. Morwitz (2014) Wood & Neal (2007)
The horizon of the Artificial Era invites comparative investigation of how goal pursuit is organized when humans and artificial systems share decision functions. For this research object, the operational unit remains an observable consumption-related goal and the sequence of authorization, information, and exchange. Proposed constructs include the correspondence between user constraints and machine recommendation, perceived agency in a delegated purchase, the stability of user authorization over time, and the extent to which automation reduces friction or generates unwanted expenditure. Their scientific status is prospective: they require carefully specified experiments, consent-aware logging, and independent replication. Existing consumer-intention findings provide an analytical starting point while the emerging empirical record must be evaluated on its own evidence. DellaVigna (2009) Sheeran & Webb (2016)
Consumer Autonomy, Privacy, and the Ethics of Conversion Metrics
Research into purchase conversion can serve distinct interests. Sellers seek transactions, whereas consumers seek outcomes that meet needs, preferences, budgets, and values. These goals can overlap, but their correspondence must be assessed rather than presumed. An intervention that increases checkout completion while also increasing unwanted subscription renewals should not be classified as a universally beneficial improvement in self-regulation. A rigorous assessment states whose welfare is considered, which outcome counts as success, and how costs, reversibility, transparency, and postpurchase consequences are measured. Consumer autonomy includes the practical ability to understand choices and refuse a transaction, particularly where an interface exploits predictable errors in attention or disclosure. DellaVigna (2009) Ajzen (1991)
Linking purchase-intention surveys to individual financial records also creates privacy and governance responsibilities. Research participants should understand what records will be connected, for what purpose, for how long, and with what protection against disclosure or unrelated use. Aggregate store sales can sometimes answer a causal question without collecting identity-level histories. When longitudinal linkage is scientifically necessary, investigators should minimize collected information, separate identifiers where feasible, and define who may access personal transaction details. These safeguards matter especially where data could reveal sensitive household circumstances or expose people to aggressive targeting. Ethical adequacy concerns the actual research procedure, not merely the presence of a statement that the dataset was anonymized. Chandon et al. (2005) Morwitz (2014)
Predictive systems can reinforce unequal commercial treatment if historical purchasing proxies for resources, location, age, or prior exclusion. Performance may differ across groups not because motivation differs but because access, visibility, and opportunity vary. Fairness evaluation should therefore examine both forecast errors and downstream allocations such as eligibility, pricing, or promotional exposure. Where automated tools act on a consumer's behalf, systems should preserve meaningful consent and a straightforward path to review or cancel transactions. These are normative design criteria informed by empirical consumer psychology; their effectiveness must be tested separately from claims about purchase-intention validity. The scientific synthesis should make these different kinds of claim explicit. DellaVigna (2009) Morwitz et al. (2007)
Critical Scientific Synthesis: A Conditional Model of Purchase Enactment
The research reviewed here supports an integrated but conditional account of consumer acquisition. Motivational processes establish why an object or outcome matters; attitudes, expected consequences, social conditions, and perceived control contribute to expressed purchasing intentions; the environment supplies or withholds options for action; and subsequent learning or preference revision can redirect the goal. Completed purchases are the observed results of these interacting processes across time. The model is a synthesis of established research traditions rather than a new validated theory, and it does not assign universal numerical weights to its components. It treats each link as a separate empirical hypothesis: motivation may predict intention, intention may forecast acquisition, and changing implementation conditions may alter the probability of completion. Fishbein & Ajzen (1975) Ajzen (1991) Bettman et al. (1998) Sheeran & Webb (2016)
This synthesis explains why the intention–behavior relation is neither trivial nor fixed. In a setting with stable preferences, familiar products, available funds, and low transaction friction, purchase intention can predict subsequent buying relatively well. When prices, available choices, household needs, or channel access change, a previously informative intention loses some of its predictive utility. Furthermore, stated intentions can be shaped by survey questions themselves and can describe social ideals or conditional hopes. Experimental interventions sometimes change intention more than behavior, showing why a forecasting measure must not be interpreted as an established mechanism of action. The strongest accounts integrate objective records with carefully timed psychological measurements and causal tests. Morwitz et al. (2007) Chandon et al. (2005) Webb & Sheeran (2006)
Evidence strength varies across conclusions. Observational associations between intention and purchase establish predictive relations under the observed measurement regime. Randomized field experiments establish causal effects of specific price, label, or planning interventions under their tested conditions. Research on self-generated validity demonstrates that measurement can sometimes participate in later behavior. Theoretical arguments concerning artificial delegation, consumer autonomy, and welfare frame questions requiring additional empirical evaluation. These strands should remain distinguishable even when they are combined in an explanatory narrative. Future progress depends on reproducible definitions of the purchase endpoint, documented opportunity constraints, uncertainty reporting, and independent replication in new markets. Chetty et al. (2009) Hainmueller et al. (2015) Chandon et al. (2005)
Open Research Questions and the Next Empirical Agenda
A central unresolved problem is identifying the circumstances under which intention changes are causally responsible for purchase changes, rather than both reflecting another factor such as price or availability. A useful program would preregister interventions designed to alter goals, information, or opportunity separately and then observe realized transactions. It would distinguish direct intervention effects from mediator claims, examine substitution and cancellation, and test whether effects persist beyond the immediate observation window. Replication across product categories would reveal which pathways generalize and which depend on a particular retail environment. The research value lies in designing discriminating tests instead of adding another broad correlation between attitudes and self-reported buying. Webb & Sheeran (2006) Sheeran & Webb (2016)
Another challenge is measuring purchase across platforms while respecting consumer privacy. Transaction data are often fragmented among retailers, payment providers, and customer devices. Cooperative designs using consented panels, aggregate treatment comparisons, or privacy-preserving linkage could improve estimation without requiring unrestricted individual surveillance. Predefined handling of off-platform purchases, delayed acquisition, returns, and repeated orders would permit comparisons across laboratories and markets. Future datasets should document base rates, follow-up losses, sample recruitment, price context, and the presence or absence of intention elicitation. This would help determine how much variation in forecast accuracy arises from psychological mechanisms and how much from the measurement system. Morwitz (2014) Kukar-Kinney et al. (2022)
Artificial recommendation and delegated purchasing create further questions about authorship of action. Researchers should test whether consumers' goals are preserved when software filters alternatives, how trust changes with system explanations, how automated reorders interact with resource constraints, and what constitutes sufficiently informed authorization. The same experiment should observe purchase completion and postpurchase satisfaction or unwanted transactions, making benefits and costs identifiable. Cross-cultural and socioeconomic comparisons should measure objective access and consumer-defined goals, not simply demographic labels. Together these directions would connect the long-standing psychology of intention and action to the transformed institutions of market exchange in the Artificial Era. Bettman et al. (1998) Ajzen (1991) DellaVigna (2009)
Conclusion
Consumer motivation, purchase intentions, and actual behavior form a connected field of inquiry whose central scientific problem is the explanation of acquisition across changing opportunities. Motivations give particular goals value; intentions express a plan, commitment, or forecast; observable transactions establish whether an acquisition occurred under defined conditions. Historical research from Fishbein and Ajzen through modern consumer-intention forecasting developed the conceptual distinction between those stages. Experimental and observational findings show that the stages are associated while remaining open to influence by product information, prices, social conditions, cognitive processes, and environmental constraints. The resulting intention–behavior gap is an empirical pattern requiring mechanism-specific explanation, not a diagnosis of the consumer. Fishbein & Ajzen (1975) Ajzen (1991) Morwitz (2014)
Research quality depends on matching what is asked to what is observed. Survey wording, timing, product and channel definitions, opportunity constraints, repeated measurement, and the coverage of transaction data all change what a reported intention can predict. Controlled studies of planning, price salience, survey reactivity, and field purchasing reveal causal pathways that correlation alone cannot establish. Stronger future studies will compare intention and transaction outcomes directly, identify the point at which plans change or are blocked, and report uncertainty and counterexamples. They will also evaluate returns, regret, substitution, and long-term consequences where immediate conversion is an incomplete measure of success. Chetty et al. (2009) Chandon et al. (2005) Gollwitzer & Sheeran (2006)
The transition From Homo to Artificial changes the technical setting of this research by introducing automated recommendation, delegated choice, and machine-mediated transactions. It does not dissolve the need for explicit operational distinctions. A user-authorized goal, an algorithmically proposed option, an executed order, and a consumer's later evaluation are separate observations and must be studied as such. The enduring contribution of consumer psychology is a disciplined account of how goals become actions through cognitive, social, material, and technological conditions. Scientific synthesis advances when those conditions are measured, mechanisms are tested, and acquisition is evaluated in relation to the purposes it serves. Bettman et al. (1998) Sheeran & Webb (2016)
References
- Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179–211. https://doi.org/10.1016/0749-5978(91)90020-T
- Armitage, C. J., & Conner, M. (2001). Efficacy of the theory of planned behaviour: A meta-analytic review. British Journal of Social Psychology, 40(4), 471–499. https://doi.org/10.1348/014466601164939
- Berridge, K. C., & Robinson, T. E. (1998). What is the role of dopamine in reward: Hedonic impact, reward learning, or incentive salience? Brain Research Reviews, 28(3), 309–369. https://doi.org/10.1016/S0165-0173(98)00019-8
- Bettman, J. R., Luce, M. F., & Payne, J. W. (1998). Constructive consumer choice processes. Journal of Consumer Research, 25(3), 187–217. https://doi.org/10.1086/209535
- Carrington, M. J., Neville, B. A., & Whitwell, G. J. (2010). Why ethical consumers don’t walk their talk: Towards a framework for understanding the gap between the ethical purchase intentions and actual buying behaviour of ethically minded consumers. Journal of Business Ethics, 97, 139–158. https://doi.org/10.1007/s10551-010-0501-6
- Chandon, P., Morwitz, V. G., & Reinartz, W. J. (2005). Do intentions really predict behavior? Self-generated validity effects in survey research. Journal of Marketing, 69(2), 1–14. https://doi.org/10.1509/jmkg.69.2.1.60755
- Chetty, R., Looney, A., & Kroft, K. (2009). Salience and taxation: Theory and evidence. American Economic Review, 99(4), 1145–1177. https://doi.org/10.1257/aer.99.4.1145
- Close, A. G., & Kukar-Kinney, M. (2010). Beyond buying: Motivations behind consumers’ online shopping cart use. Journal of Business Research, 63(9–10), 986–992. https://doi.org/10.1016/j.jbusres.2009.01.022
- Deci, E. L., & Ryan, R. M. (2000). The “what” and “why” of goal pursuits: Human needs and the self-determination of behavior. Psychological Inquiry, 11(4), 227–268. https://doi.org/10.1207/S15327965PLI1104_01
- DellaVigna, S. (2009). Psychology and economics: Evidence from the field. Journal of Economic Literature, 47(2), 315–372. https://doi.org/10.1257/jel.47.2.315
- Dhar, R. (1997). Consumer preference for a no-choice option. Journal of Consumer Research, 24(2), 215–231. https://doi.org/10.1086/209506
- Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Addison-Wesley. https://people.umass.edu/aizen/f%26a1975.html
- Gollwitzer, P. M., & Sheeran, P. (2006). Implementation intentions and goal achievement: A meta-analysis of effects and processes. Advances in Experimental Social Psychology, 38, 69–119. https://doi.org/10.1016/S0065-2601(06)38002-1
- Hainmueller, J., Hiscox, M. J., & Sequeira, S. (2015). Consumer demand for fair trade: Evidence from a multistore field experiment. Review of Economics and Statistics, 97(2), 242–256. https://doi.org/10.1162/REST_a_00467
- Johnson, E. J., & Goldstein, D. (2003). Do defaults save lives? Science, 302(5649), 1338–1339. https://doi.org/10.1126/science.1091721
- Kukar-Kinney, M., & Close, A. G. (2010). The determinants of consumers’ online shopping cart abandonment. Journal of the Academy of Marketing Science, 38(2), 240–250. https://doi.org/10.1007/s11747-009-0141-5
- Kukar-Kinney, M., Scheinbaum, A. C., Orimoloye, L. O., Carlson, J. R., & He, H. (2022). A model of online shopping cart abandonment: Evidence from e-tail clickstream data. Journal of the Academy of Marketing Science, 50(5), 961–980. https://doi.org/10.1007/s11747-022-00857-8
- Laibson, D. (1997). Golden eggs and hyperbolic discounting. Quarterly Journal of Economics, 112(2), 443–478. https://doi.org/10.1162/003355397555253
- Milkman, K. L., Beshears, J., Choi, J. J., Laibson, D., & Madrian, B. C. (2011). Using implementation intentions prompts to enhance influenza vaccination rates. Proceedings of the National Academy of Sciences, 108(26), 10415–10420. https://doi.org/10.1073/pnas.1103170108
- Morwitz, V. (2014). Consumers’ purchase intentions and their behavior. Foundations and Trends in Marketing, 7(3), 181–230. https://doi.org/10.1561/1700000036
- Morwitz, V. G., Johnson, E. J., & Schmittlein, D. C. (1993). Does measuring intent change behavior? Journal of Consumer Research, 20(1), 46–61. https://doi.org/10.1086/209332
- Morwitz, V. G., Steckel, J. H., & Gupta, A. (2007). When do purchase intentions predict sales? International Journal of Forecasting, 23(3), 347–364. https://doi.org/10.1016/j.ijforecast.2007.05.015
- Ouellette, J. A., & Wood, W. (1998). Habit and intention in everyday life: The multiple processes by which past behavior predicts future behavior. Psychological Bulletin, 124(1), 54–74. https://doi.org/10.1037/0033-2909.124.1.54
- Sheeran, P. (2002). Intention–behavior relations: A conceptual and empirical review. European Review of Social Psychology, 12(1), 1–36. https://doi.org/10.1080/14792772143000003
- Sheeran, P., & Webb, T. L. (2016). The intention–behavior gap. Social and Personality Psychology Compass, 10(9), 503–518. https://doi.org/10.1111/spc3.12265
- Sniehotta, F. F., Presseau, J., & Araújo-Soares, V. (2014). Time to retire the theory of planned behaviour. Health Psychology Review, 8(1), 1–7. https://doi.org/10.1080/17437199.2013.869710
- Webb, T. L., & Sheeran, P. (2006). Does changing behavioral intentions engender behavior change? A meta-analysis of the experimental evidence. Psychological Bulletin, 132(2), 249–268. https://doi.org/10.1037/0033-2909.132.2.249
- Wood, W., & Neal, D. T. (2007). A new look at habits and the habit–goal interface. Psychological Review, 114(4), 843–863. https://doi.org/10.1037/0033-295X.114.4.843
Social Influence, Approval, and Consumer Identity in Action
Purchasing can be publicly expressive even when the transaction is private. Consumers anticipate reactions from family, peers, communities, or professional networks, and these expectations can shape which products they consider and how confidently they state their intentions. The subjective-norm component in planned behavior offers one operational route to studying perceived approval, though norm measures require explicit specification of the relevant people and the behavior at issue. Social expectations can affect intention formation while purchase opportunities depend on household budgets, relationship obligations, or restrictions. The strongest design separates individual evaluation, anticipated social judgment, and actual observed acquisition rather than collapsing social and private motives into one general favorable attitude. Ajzen (1991) Armitage & Conner (2001)
Self-determination theory adds questions about the quality of motivation and the social conditions under which goals are pursued. A purchase can support personally endorsed interests, a practical aspiration, or a wish for belonging. The outward similarity of two products reveals little about the underlying regulation. Consequently a researcher interested in identity-driven purchase behavior must measure relevant values and self-construal rather than infer motives from spending alone. Social approval may also influence self-report: survey respondents can describe what a conscientious or admired consumer ought to buy instead of predicting their own likely spending. Separating normative ideals from forecasts becomes especially important for ethically charged or visibly status-related categories. Deci & Ryan (2000) Carrington et al. (2010)
Retail experiments suggest that attitudes toward product attributes can have behavioral effects conditional on price. Hainmueller, Hiscox, and Sequeira studied fair-trade labeling in grocery stores and observed changes in coffee sales alongside heterogeneous price sensitivity. This design directly observes a market outcome rather than relying entirely on a declaration of ethical purchase intent. The inference is still context-bound: the intervention concerns a particular label, products, stores, and consumer populations, and it does not establish an invariant social motive for every purchaser. Field evidence is especially valuable when social ideals and budget constraints are likely to diverge. Sustainable consumption and the specialized ethical intention–action gap remain the subjects of their separate dedicated dossier. Hainmueller et al. (2015) Carrington et al. (2010)