Kano Model
1. Overview
A. Definition
The Kano Model is a quality-classification and prioritization technique that analyzes the quality attributes of a product or service as a nonlinear relationship between two axes — "the functional level achieved (physical fulfillment) when the attribute is provided" and "the resulting customer satisfaction" — and classifies each attribute into five categories: attractive, one-dimensional, must-be, indifferent, and reverse quality, thereby distinguishing "which requirements lift satisfaction from which requirements prevent dissatisfaction."
The Kano Model was presented in the 1984 paper "Attractive quality and must-be quality" by Professor Noriaki Kano of Japan. Its core insight is that "quality and satisfaction are not proportional." Traditional quality thinking assumed that the more and better the functionality provided, the more linearly customer satisfaction rises; in reality, however, there coexist attributes that are taken for granted and barely raise satisfaction no matter how well provided (e.g., a smartphone's call connectivity) and attributes that delight even when provided only a little (e.g., unexpected convenience features). Kano systematized through a two-dimensional graph the point that the very shape of the fulfillment-satisfaction relationship differs by attribute.
The properties of the Kano Model can be summarized in four points. First, it treats satisfaction and dissatisfaction as separate dimensions — removing dissatisfaction does not create satisfaction (it aligns with Herzberg's motivation-hygiene theory). Second, it treats quality attributes asymmetrically as "nice to have" versus "must not be missing." Third, the classification result becomes a basis for requirements prioritization, supporting decisions on allocating limited development resources. Fourth, an attribute's category is not fixed but shifts over time and across customer segments (dynamic characteristic). These four properties make the Kano Model not a mere survey tool but an analytical framework for product strategy.
B. Background and Necessity
The quality movement in Japanese manufacturing of the 1970s–80s stood on the linear assumption that "reducing defects improves quality, and improving quality satisfies customers." Yet even after lowering defect rates to the extreme, a stagnation of satisfaction was observed in which customer satisfaction no longer rose. Zero defects only created "a state without dissatisfaction," not "delight." Firms had to answer "we raised quality further, so why is the market response flat?", and the Kano Model emerged as a tool that distinguishes "which quality must be raised for satisfaction to actually rise."
In software and IT services this necessity grew even larger. Requirements always exceed resources and schedule, and not every feature can be implemented equally. Putting in "everything the customer said they wanted" wastes resources on indifferent features that do not contribute to satisfaction, while missing essential must-be features invites fatal dissatisfaction. Thus in [[requirements-engineering]], a step that "after collecting requirements, classifies and prioritizes them by how each requirement contributes to satisfaction" became essential, and the Kano Model, alongside prioritization techniques such as MoSCoW and AHP, took its place as a core means.
This necessity appears identically in building public and enterprise information systems. Informatization projects often collect hundreds of user requirements, but reflecting all of them with equal weight easily yields a "feature-rich yet vaguely useful" system that overruns budget and schedule while delivering low perceived satisfaction. Here the Kano Model separates "must-be requirements that absolutely must be met (e.g., compliance with statutory processing deadlines, data consistency)" from "attractive requirements that lift satisfaction (e.g., intelligent autocomplete, proactive notifications)," providing grounds for dividing a limited budget into "investment to avoid failure" and "investment to stand out." This shifts requirement prioritization from subjective consensus to data-driven decision-making.
Understanding why one should not "implement stated requirements as is" makes Kano's value clear. Customers do not voice must-be quality because it is taken for granted (it is easily omitted from requirements specifications), and they cannot request attractive quality because they cannot imagine it (so it does not appear in a survey's "features you want"). That is, chasing only customers' explicit statements falls into the double trap of omitting must-be quality and failing to discover attractive quality. The Kano survey is fundamentally different from a simple preference survey in that it reveals these two hard-to-ask-about categories through indirect question pairs.
C. Characteristics
The greatest characteristic of the Kano Model is that it captures the asymmetry of satisfaction-dissatisfaction quantitatively and visually. Must-be quality keeps satisfaction at 0 even when fulfilled but plunges satisfaction when unfulfilled, whereas attractive quality causes no dissatisfaction when unfulfilled but spikes satisfaction when fulfilled. This asymmetric curve directly provides the strategic implication that "resources for defense (blocking dissatisfaction) and resources for offense (differentiation) must be allocated differently."
The second characteristic is the design that classifies attributes through indirect question pairs (functional + dysfunctional). Respondents are asked in pairs "how do you feel if this feature is present" and "how do you feel if this feature is absent," and the category is inferred from the combination of the two answers. This bypasses the psychology whereby a respondent finds it hard to say "this is taken for granted," filling the gap between statements and actual reactions. In this respect the Kano Model functions as a complementary technique that systematically surfaces implicit requirements that [[requirements-specification]] tends to miss.
The third characteristic is the practical orientation whereby classification connects directly to a resource-allocation strategy. Whereas other requirement-analysis techniques only rank "what is important," Kano gives the causal shape of "in which direction and by how much does satisfaction move if this requirement is met or unmet," so attributes can be mapped directly to distinct investment purposes: defense (blocking dissatisfaction), competition (performance edge), and offense (differentiation). This shift of perspective — seeing quality not as the one dimension of "more or less" but as "several curves of different kinds" — is the Kano Model's essential contribution.
2. The Five Categories of Quality Elements and the Satisfaction-Fulfillment Curve
The backbone of the Kano Model is drawing, on a two-dimensional plane with the horizontal axis as "the attribute's physical fulfillment (how well the function was provided)" and the vertical axis as "customer satisfaction," a differently shaped curve for each attribute. The shape of the curve is the category, and the conceptual diagram below shows the curves of the five categories and the arrangement of the satisfaction/dissatisfaction regions.
flowchart TB
subgraph PLANE["Kano 2D plane (x: fulfillment, y: satisfaction)"]
A["Attractive quality<br/>no dissatisfaction even if unfulfilled → satisfaction spikes if fulfilled"]
O["One-dimensional quality<br/>satisfaction rises linearly with fulfillment"]
M["Must-be quality<br/>satisfaction 0 even if fulfilled → dissatisfaction plunges if unfulfilled"]
I["Indifferent quality<br/>fulfillment unrelated to satisfaction"]
R["Reverse quality<br/>more provision → more dissatisfaction"]
end
A --> DIFF["Differentiation/delight (offense resources)"]
O --> COMP["Competitiveness metric (performance competition)"]
M --> BASE["Entry qualification/dissatisfaction defense (defense resources)"]
I --> WASTE["Over-investment risk (resource saving)"]
R --> REMOVE["Target for removal/making optional"]
A. Must-be quality (Must-be, M). Basic attributes the customer presupposes "must obviously be there." Even when fulfilled, satisfaction stays neutral (0), but the slightest shortfall makes satisfaction drop sharply. By analogy it is an entry ticket — you are not praised for having it, but without it you are eliminated. A banking app's "accurate balance display," a car's "braking system," and a cloud service's "no data loss" are typical. Must-be quality is easily omitted from requirement collection because customers do not mention it, and if even one collapses it overwhelms all other attractive elements and ruins the overall evaluation. Therefore must-be quality aims at "perfect fulfillment" and is not a target of over-investment (because beyond a certain level satisfaction no longer rises).
B. One-dimensional quality (One-dimensional, O). An attribute where the higher the fulfillment, the more satisfaction rises linearly in proportion, and where dissatisfaction grows linearly when unfulfilled. Often called performance quality, it is the domain customers explicitly request and directly compare against competitors. A smartphone's battery life, a web service's response speed, and a storage device's capacity/price ratio are examples. Since "more is better" holds for one-dimensional quality, it becomes the main battlefield of competition, and an edge here leads directly to market share. However, when all competitors compete on the same axis, they easily fall into diminishing marginal utility and bleeding competition, so strategically one must also seek a pivot of the axis toward attractive quality.
C. Attractive quality (Attractive, A). An attribute that causes no dissatisfaction when absent but delivers beyond-expectation delight when present. A domain customers did not request because they could not even imagine it, it is the source of differentiation and innovation. The early iPhone's multitouch, Tesla's OTA updates, and the generative-AI assistants embedded in 2023–2024 productivity tools are examples. Because attractive quality carries no penalty for being unfulfilled, it has the asymmetric payoff structure of "no loss if omitted, but a decisive edge if included." From a professional engineer's perspective, discovering attractive quality has limits with surveys alone, so it must be combined with exploratory methods such as ethnographic observation, latent-needs analysis, and prototyping.
D. Indifferent and reverse quality (Indifferent I, Reverse R). Indifferent quality is an attribute that has no effect on satisfaction whether present or not, and spending resources on it is pure waste. Reverse quality is an attribute where more provision increases dissatisfaction, such as excessive notifications, forced sign-up, and needlessly complex features. Identifying these two is as important for deciding "what to remove or make optional" as for "what to include," and it provides grounds to reduce the [[technical-debt]] and maintenance cost caused by feature creep. In addition, when survey responses are contradictory, they are classified as Questionable (Q) to check response reliability.
The knack for reading the five categories in one picture is to see whether the curve passes through the origin and which quadrant it straddles. One-dimensional quality is an upward-right line through the origin; must-be quality has a steep slope only in the lower-left (unfulfilled/dissatisfaction) and flattens on the right (fulfillment). Attractive quality, conversely, is flat on the left and spikes in the upper-right (fulfillment/satisfaction). Indifferent quality is a horizontal line, and reverse quality a downward-right line. Once trained to translate "shape of curve = strategic meaning," one can, even without a survey, hypothesize which category to aim a new feature at during the design stage and then verify that hypothesis with a survey — an efficient cycle.
| Category | When fulfilled | When unfulfilled | Strategic role | Resource allocation |
|---|---|---|---|---|
| Must-be (M) | Satisfaction 0 (neutral) | Dissatisfaction plunge | Entry qualification/defense | Only up to full fulfillment |
| One-dimensional (O) | Satisfaction linear↑ | Dissatisfaction linear↓ | Performance competitiveness | Competitive level+α |
| Attractive (A) | Satisfaction spike | No dissatisfaction | Differentiation/delight | Selective focused investment |
| Indifferent (I) | No effect | No effect | None | Avoid investment |
| Reverse (R) | Dissatisfaction rises | (N/A) | Removal/making optional | Negative (-) investment |
3. The Kano Survey and Analysis Process
Kano classification is done not by intuition but by structured surveys and aggregation. The key is to ask, for each attribute, a functional question (how do you feel if the feature is present) and a dysfunctional question (how do you feel if the feature is absent) in pairs, and to judge the category by applying the combination of the two responses to an evaluation table. The flow below is the typical procedure from attribute definition to deriving priorities.
flowchart LR
DEF["1. Define target attributes"] --> Q["2. Design functional/dysfunctional question pairs (5-point scale)"]
Q --> SURVEY["3. Collect survey (target customer group)"]
SURVEY --> TABLE["4. Classify response pairs via Kano table (A/O/M/I/R/Q)"]
TABLE --> MODE["5. Determine the mode category per attribute"]
MODE --> CS["6. Compute Better/Worse coefficients"]
CS --> PRI["7. Reflect into prioritization/roadmap"]
PRI -.feedback.-> DEF
Functional/dysfunctional question pairs and scale. Each question is usually answered on a 5-point scale: "① I like it ② It is a must ③ I am neutral ④ I can live with it ⑤ I dislike it." For example, "What if the app supports dark mode?" (functional) and "What if the app does not support dark mode?" (dysfunctional) are asked in pairs. This question pair elicits both positive and negative reactions from the respondent, capturing the asymmetry that does not surface when only one side is asked. If a feature is described too leadingly in question design, it is over-classified as attractive, so neutral and concrete wording is important.
Kano evaluation table (classification matrix). Applying the two responses to a 5×5 grid maps each response pair to a single category. The table below is the core part of the standard evaluation table, where the combination of functional (rows) and dysfunctional (columns) determines the category.
| Functional\Dysfunctional | Like | Must-be | Neutral | Live-with | Dislike |
|---|---|---|---|---|---|
| Like | Q | A | A | A | O |
| Must-be | R | I | I | I | M |
| Neutral | R | I | I | I | M |
| Live-with | R | I | I | I | M |
| Dislike | R | R | R | R | Q |
For example, answering an attribute "functional=Like, dysfunctional=Dislike" classifies it as the one-dimensional (O) indicated by the table's (Like, Dislike) cell. "Good to have and bad to miss" matches exactly the definition of one-dimensional quality — "satisfaction changes in proportion to fulfillment." Conversely, "functional=Like, dysfunctional=Must-be" is attractive (A) — an asymmetric reaction of being pleased when present but taking it for granted when absent. One aggregates all respondents and takes the per-attribute mode category as that attribute's classification, but when responses are dispersed across categories one suspects the category may differ by customer segment.
Practical cautions on survey design and sample. Since respondent fatigue rises sharply as question pairs increase in a Kano survey, the attributes covered in one survey are usually limited to within 15–30, with important candidates pre-selected. Secure a sample large enough to draw a meaningful proportion per category (in practice, tens to hundreds per segment), and collect respondent attributes (proficiency, usage frequency, pricing plan) together so segments do not mix, preparing for cross-analysis. Also, describing "how do you feel if the feature is present" too attractively causes framing bias that skews responses toward A, so neutralize the question wording, and to check response consistency, deliberately mix re-questions into some attributes and monitor the Q (questionable) ratio. Analysis results should present not only the mode but the full category distribution together with the Better/Worse coefficients, to prevent the error of "lumping everything into a single mode category."
Better/Worse coefficients (customer satisfaction coefficients). Since the mode category alone makes it hard to compare the weight among attributes, the two coefficients proposed by Timko are used together. Better (satisfaction increment index) = (A+O) / (A+O+M+I) represents the potential for satisfaction to rise when the feature is provided, and Worse (dissatisfaction index) = -(O+M) / (A+O+M+I) represents the risk of growing dissatisfaction when not provided (each symbol is the proportion responding in that category). For example, if among 100 people A=40, O=30, M=20, I=10, then Better=(40+30)/100=0.70 and Worse=-(30+20)/100=-0.50. A large Better marks a differentiation-investment candidate, a large absolute Worse marks a defense line that must be held, and placing the two coefficients on a 2D map reveals the per-attribute investment strategy at a glance.
This Better-Worse map is read as four regions. If both Better and |Worse| are high, the one-dimensional character is strong, so it is "the core axis of competition" and must receive continued investment; if only Better is high, it is an attractive attribute, a "differentiation opportunity" for selective focus. If only |Worse| is high, it is a must-be attribute, "a baseline to absolutely hold"; if both are low, it is an indifferent attribute, a target for "avoiding investment." Thus the two coefficients turn the qualitative classification of categories into continuous coordinates, enabling prioritization even within the same category and backing quarterly investment weights in the roadmap with numbers. However, since the coefficients are sensitive to sample and items, report them with confidence intervals and exercise the restraint of treating them as a relative-comparison tool, not an absolute criterion.
4. Comparison and Application Cases
Comparison with traditional quality thinking and satisfaction surveys. A simple satisfaction survey asks "how satisfied are you with this feature" in only one direction, failing to distinguish must-be quality (already fulfilled so satisfaction is neutral) from indifferent quality (no effect to begin with). Both appear as "satisfaction average." The Kano Model adds the dysfunctional question to separate "does it burst into dissatisfaction if absent (M)" from "is it fine even if absent (I)." This difference connects directly to resource allocation — the former is a defense line that must be held, the latter a target to cut boldly. That is, Kano's added value lies in revealing that "even if the apparent satisfaction in the fulfilled state is the same, the strategic meaning of the attribute is entirely different."
In addition, it is worth noting as a comparison that the Kano Model conceptually resembles Herzberg's motivation-hygiene theory. Hygiene factors (dissatisfaction if absent, no satisfaction even if present) correspond to must-be quality, and motivation factors (lift satisfaction if present) correspond to attractive quality. Both theories share the insight that "satisfaction and dissatisfaction are not two ends of the same axis but different axes." This similarity shows that the Kano Model is not merely a survey technique but stands on a general theory of human satisfaction psychology, and therefore transfers broadly to the user-experience design of not only products but also internal systems and organizational services.
Relationship with IPA and QFD. Importance-Performance Analysis (IPA) arranges an attribute's importance and current performance on a plane to find improvement priorities, but it relies on respondents' direct evaluation of "importance." Yet attractive quality is easily underrated in IPA because respondents "do not say" it is important. Placing the Kano Model ahead of IPA to first identify attractive elements can complement this blind spot. Also, adjusting the weight of customer requirements by Kano category (weight correction for must-be and attractive elements) in the House of Quality (HoQ) of Quality Function Deployment (QFD, which links with [[requirements-engineering]]) raises the accuracy of translating the voice of the customer into design characteristics.
Finally, mutual complementarity with AHP and weighted scoring is also important in practice. If Kano reveals "what kind of satisfaction an attribute contributes to" (qualitative nature), AHP, used together in [[requirements-engineering]], computes "relative importance among attributes" (quantitative weights) via pairwise comparison. Combining the two techniques enables two-stage prioritization: first filter out must-be, reverse, and indifferent elements with Kano to compress the candidates, then assign AHP weights to the remaining attractive and one-dimensional elements to set the development order. Thus Kano is strongest when combined as a preprocessing filter for other prioritization techniques rather than completed on its own.
Concrete cases. ① When a fintech app ran a Kano survey (n=500) on 20 kinds of transfer features, "biometric-authentication login" was classified as M (Worse=-0.72), revealing a defense line that triggers "distrust if absent," while "AI spending-analysis report" became a differentiation candidate as A (Better=0.68). The team separated biometric authentication as the top stabilization task and the AI report as an innovation task to showcase front and center in marketing on the next quarter's roadmap. ② In the automotive industry, navigation and rear cameras, which were attractive quality in the 1990s, shifted to one-dimensional and must-be quality in the 2020s, so vehicles without them came to receive a near-reverse-quality deduction in used prices. ③ When a SaaS firm hid many advanced-setting features classified as indifferent (I) from the default UI and moved them to an "advanced mode" option, the new-user onboarding completion rate rose significantly (the effect of removing reverse and indifferent quality). ④ When a public civil-complaint portal classified "24-hour processing-status notification" as A (Better=0.61) and "a variety of screen themes" as I via a Kano survey, and concentrated its limited budget on enhancing notifications while deferring the theme feature, both the return-visit rate and complaint-handling satisfaction improved. What these cases share is that dividing "what to include" from "what to cut" by data changes perceived satisfaction greatly even on the same budget.
5. Deep Dive — Dynamic Kano and the Evolution of Data-Driven Analysis
The most important deep-dive topic of the Kano Model is its dynamic characteristic (dynamic Kano). The category of a quality attribute is not fixed but degenerates (becomes obsolete) over time in the direction attractive → one-dimensional → must-be. An innovation that at first delighted descends to the performance-competition axis (one-dimensional) once competitors imitate it and it becomes common, and further becomes "something that must obviously be there" (must-be). The smartphone camera shifting from an early attractive element to today's must-be element is representative. Since this dynamic means "today's differentiation is tomorrow's basic," firms must continuously rediscover attractive quality and must run the Kano survey periodically and repeatedly, not once, to track category shifts. From a professional engineer's perspective, this gives the strategic implication that a "differentiation pipeline" must be constantly maintained in the product roadmap and [[okr]]/portfolio management.
The second direction of evolution is data- and AI-driven Kano analysis. Traditional surveys have the limits of high cost and time and small samples. Recently, applying [[text-opinion-mining]] and sentiment analysis to large volumes of unstructured text such as app reviews, social media, and customer inquiries, the practice and research of automatically estimating Kano categories by how a feature mention combines with positive/negative sentiment are growing. For example, a feature with a strong negative context of "inconvenient because it is missing" is estimated as must-be, and a feature with a strong positive context of "surprised that it is there" as attractive. This enables continuous Kano monitoring that constantly reflects millions of real-usage feedback items, complementing the weakness of the slow survey cycle.
When applying data-driven analysis, one must be careful about bias correction. Review and inquiry text tends to over-sample dissatisfied users, so must-be quality (dissatisfaction context) may be inflated and attractive quality (quiet satisfaction) underrated. Therefore it is safer to cross-validate text-based estimates with traditional surveys and to convert into categories after correcting the base rate of positive/negative ratios. Triangulating the qualitative (survey) and the quantitative (large-scale text) this way offsets each method's sampling bias against the other and reaches a more robust classification.
The third is combination with agile and product-management methodologies. Kano categories are used directly in backlog prioritization — must-be quality is necessarily included in the MVP (minimum viable product), attractive quality is placed into differentiation sprints, and indifferent and reverse quality are removed from the backlog or sent back to the discovery stage. Some organizations attach Kano tags to each user story to distinguish "for defense/for competition/for delight," linking this to value-hypothesis validation (A/B testing) from a [[data-product]] perspective to verify the validity of the classification with real data. That is, the Kano Model serves as the entrance to a loop that closes from qualitative classification into quantitative experiment.
The practical benefit of this combination lies in unifying the language of decision-making. Planning, development, and marketing often argue "this feature is important" on different grounds, but Kano categories and Better/Worse coefficients converge the debate into a common quantitative language, such as "this is a defense line of Worse=-0.7, so deferring it invites churn" or "this is a differentiation of Better=0.6, so it becomes the core of the marketing message." With this, a backlog-prioritization meeting turns from a contest of the loudest person's intuition into a consensus process grounded in customer data. From a professional engineer's perspective, institutionalizing this cycle of "classification → coefficient → experiment → reclassification" into the organization's requirements-management process is the key task.
6. Considerations and Implications
First, combining analysis scope and method (application strategy). The Kano Model is strong at classifying "attributes already thought of," but cannot create attractive elements not yet imagined. Therefore it is desirable to place observation, ethnography, and prototyping that discover latent needs at the front, and to position Kano as the back-end tool that validates and prioritizes their results — a dual structure. Expecting innovation from surveys alone over-interprets the Kano Model's role.
Second, segment dependence and the trap of aggregation (trade-off). Even the same attribute may fall into different categories by customer group (novice/expert, price-sensitive/premium). Aggregating all responses into a single mode buries inter-segment conflicts in the average, risking the design of a product "optimal for no one." Therefore, when category responses are dispersed, analyze by segment separately, specify the target segment, and then set priorities. The balance between the convenience of aggregation and segment precision is the crux.
Third, dynamic tracking and reinvestment (outlook). Since the mustbe-ization of attractive quality is inevitable, Kano analysis is meaningful only when repeated and made continuous. A roadmap decided by a one-off analysis fails to reflect category shifts and ages quickly. An organization should update Kano metrics (category, Better/Worse) on a quarterly/semiannual basis and operate a quality-attribute portfolio that manages the life cycle of differentiation elements. The outlook is that real-time feedback data combined with AI estimation will make an "always-on dynamic Kano dashboard" the standard instrument panel of product operations.
Fourth, the limits of innovation creation and its complement (methodology combination). Because the Kano survey relies on "already-presented features" that respondents can evaluate, it cannot itself create new attractive elements not present in the market. That is, Kano is a tool to validate and classify innovation, not to conceive it. Therefore it is desirable to combine it with hypothesis-setting from design thinking and lean startup and with latent-needs observation to first generate "candidate ideas," and to use Kano to measure the satisfaction-contribution structure of those candidates — a generation-validation division of labor. Making this boundary clear avoids the overconfidence that "just run a survey and innovation comes out."
Fifth, contextualizing quantitative metrics and governance (linkage/operation). Better/Worse coefficients or the mode category do not by themselves automate decisions. They must be interpreted in combination with context such as strategy, brand, regulation, and [[quality-cost]], and among must-be quality, items that can never be conceded — such as safety, security, and regulatory compliance — even when the Worse coefficient comes out low, must be protected by separate governance. Link Kano metrics to the organization's requirement-prioritization system and release criteria, but maintaining the contextual judgment of "why this category" rather than the mechanical application of numbers is professional-engineer-level use.
In sum, the Kano Model breaks the simple equation "quality equals satisfaction" and redefines quality as a bundle of several curves with different purposes. The capability required of a professional engineer is not to memorize this classification but to discern by data which attributes are defense lines and which are differentiation levers in the context of a given product, service, or information system, and to institutionalize that discernment into requirements-management, roadmap, and governance processes. When it manages not only static classification but also the temporal shift of categories, the Kano Model becomes, beyond a one-off survey tool, a compass for a sustained quality strategy.
References
- Kano, N. et al., "Attractive Quality and Must-Be Quality" (1984), Journal of the Japanese Society for Quality Control
- Timko, M., "An Experiment in Continuous Analysis" (Kano's Methods for Understanding Customer-Defined Quality)
- Interaction Design Foundation — Kano Model overview: https://www.interaction-design.org/literature/topics/kano-model
In one line: The Kano Model analyzes quality attributes as a nonlinear fulfillment-satisfaction relationship, classifies them into must-be, one-dimensional, attractive, indifferent, and reverse quality, and uses functional/dysfunctional question pairs and Better/Worse coefficients to distinguish "defense elements that prevent dissatisfaction from differentiation elements that lift satisfaction," supporting requirements prioritization.