Not All Course Attributes Are Equal: The Kano Model and the Asymmetry of Student Satisfaction

In brief

The arithmetic mean of a course-evaluation scale quietly assumes something false: that every attribute moves satisfaction in a straight, symmetric line, so a one-point gain anywhere is worth the same. The Kano model (Kano, Seraku, Takahashi & Tsuji, 1984), rooted in Herzberg's two-factor theory of motivation (Herzberg, Mausner & Snyderman, 1959), shows satisfaction is asymmetric. Some course attributes are must-be — their absence infuriates students but their presence earns no credit (a working VLE, timely marks). Others are attractive — their presence delights but their absence is barely noticed (an inspiring guest lecture). A few are one-dimensional — more is linearly better. Averaging collapses these three fundamentally different curves into one number, and so hides the very information a quality office needs: fixing a must-be prevents dissatisfaction, while adding a delighter creates satisfaction, and the two are not substitutes.

---

What the research says

Frederick Herzberg's two-factor theory (Herzberg et al., 1959) broke a long-standing assumption: that satisfaction and dissatisfaction are opposite ends of one scale. Herzberg argued they are distinct dimensions driven by different factors — "hygiene" factors cause dissatisfaction when absent but do not motivate when present, while "motivators" create satisfaction. Remove the assumption of a single continuum and much of satisfaction research changes shape.

Noriaki Kano and colleagues operationalised this for quality management (Kano et al., 1984). Their model classifies each attribute of a product or service into categories by how its presence and absence map to satisfaction:

• Must-be (basic) quality — expected; taken for granted. Present, it does nothing for satisfaction; absent, it causes strong dissatisfaction. (In a course: functioning learning platform, accurate timetable, marks returned on time.)
• One-dimensional (performance) quality — satisfaction rises roughly linearly with performance. More is better, less is worse. (Clarity of explanation; quality of feedback.)
• Attractive (excitement) quality — unexpected. Present, it delights; absent, no penalty. Kano's central insight is that these have "the most decisive influence on user satisfaction but their absence does not generate dissatisfaction." (An unexpected industry guest, an especially engaging optional resource.)
• Indifferent — students do not care either way.
• Reverse — the attribute actively displeases some students when present.

Kano's elicitation method is distinctive: each attribute is probed with a functional / dysfunctional question pair — "How would you feel if the course had X?" and "How would you feel if it did not?" — and the pattern of the two answers classifies the attribute. This pairing is what makes the asymmetry visible; a single "how satisfied" item cannot.

The model has a solid, growing base in higher education. Madzík, Budaj, Mikuláš and Zimon (2019), in Administrative Sciences, ran a pilot applying the Kano model to student requirements and found attributes such as "practice orientation" and "quality resources" to be the most stable requirements, while others were far less stable in their classification. Broader reviews of Kano in education (e.g., the 2024 systematic review of Kano applications in education) document its use for course design, student services and campus experience, repeatedly finding that academic-support items behave as must-be or one-dimensional while enrichment items behave as attractive.

---

Why it matters for course evaluation in practice

Once you accept asymmetry, several standard QA habits look mistaken.

A high mean can hide a broken must-be. Suppose a module scores 4.3 overall but a fifth of students report marks were returned late. In a linear model the late feedback is "averaged away." In a Kano frame it is a must-be failure — the single most corrosive kind, because failing a basic expectation drives dissatisfaction disproportionately and poisons the whole evaluation. The mean tells you the course is fine; Kano tells you it has a leak below the waterline.

Chasing delighters cannot rescue a broken basic. A common instinct after a mediocre evaluation is to add something exciting — a novel assignment, a flashy guest. But you cannot compensate for a missing must-be with an attractive attribute; students do not trade a working timetable for a guest lecture. Kano tells you to fix must-bes first, then invest in delighters — the opposite of the "add something shiny" reflex.

Delighters decay. Today's attractive attribute becomes tomorrow's expectation. An innovation that once delighted students migrates, over cohorts, into a must-be. A quality cycle therefore needs to re-classify attributes periodically, not assume last year's map holds.

It reframes benchmarking. Comparing two courses on a single mean is misleading when one is strong on must-bes (dependable but unremarkable) and another is strong on delighters (exciting but occasionally drops a basic). They are different risk profiles, not different points on one line.

---

Limitations and honest caveats

The Kano model is illuminating, not infallible, and a rigorous reader should weigh its weaknesses.

Classification is unstable and sample-dependent. As Madzík et al. (2019) themselves found, some attributes classify inconsistently across respondents and subgroups. The standard "most frequent category wins" rule can mask genuine heterogeneity — an attribute that is must-be for one cohort and attractive for another is not well described by a single label.

The functional/dysfunctional pairing is demanding. Kano doubles the number of questions per attribute and asks students a slightly unnatural counterfactual ("how would you feel if it did not have this?"). Fatigue and confusion are real risks, and forced-choice Kano questionnaires can be cognitively heavy.

It is categorical, not continuous. Classic Kano assigns a category, not a magnitude. Extensions (the Better-Worse / CS-coefficient index, penalty-reward analysis) add quantification but move away from the model's original simplicity and reintroduce assumptions.

Cultural and disciplinary variation. What counts as a basic expectation differs across national systems, disciplines and student populations. A Kano map built in one context should not be transported wholesale to another — the same caution that applies to cross-cultural response styles applies here.

It describes preferences, not learning. Kano is a satisfaction model. A delighter that students love may not improve learning outcomes, and a must-be they take for granted (rigorous assessment) may matter enormously for learning. Kano must be triangulated with outcome evidence, never used alone to redesign a course.

---

How Koji incorporates this

The asymmetry Kano describes is invisible to any instrument that only asks "how satisfied are you?" Koji for Education is designed to surface it rather than average it away.

• Paired probing, conversationally. Kano's functional/dysfunctional logic maps naturally onto Koji's AI-moderated conversational interviews: rather than firing two stiff counterfactual items per attribute, the moderator can explore how a student feels about a feature's presence and its absence in natural dialogue, then follow up — capturing the classification signal with far less respondent fatigue than a rigid Kano grid.
• Structured items for the scaffold. Where a structured instrument is wanted, Koji's yes_no, single_choice and scale question types support the functional/dysfunctional pairing, and ranking items help separate must-be basics from attractive extras.
• Asymmetry from the open text. Koji's automatic thematic analysis reads the qualitative feedback for the tell-tale signature of each category: must-be attributes appear in comments almost only when they fail (nobody praises a working timetable, everyone complains about a broken one), while attractive attributes appear almost only when present. That sentiment-asymmetry is precisely what distinguishes a basic from a delighter, and it is designed to be extracted automatically rather than hand-coded.
• Guarding against the wrong fix. By flagging must-be failures distinctly from missing delighters, Koji's bias-aware reporting is designed to steer a QA office toward fixing basics before chasing excitement — the priority order Kano prescribes and a mean obscures.

Koji does not claim to eliminate the instability inherent in Kano classification — it is designed to mitigate the fatigue and detection problems and to add the qualitative evidence categorical labels lack. Institutions extending this thinking to product and customer research use the same conversational engine on Koji's core platform at koji.so.

---

Watching a delighter decay

The migration of attractive attributes into must-be expectations is not hypothetical, and a quality cycle should track it. When recorded lectures first appeared they were an attractive attribute — their presence delighted students, their absence went unremarked. Within a few cohorts they became a must-be: students now barely mention recordings when they exist but react sharply when a session is not captured. An institution that classified recordings as a delighter three years ago and never re-ran the analysis will systematically under-invest in what has quietly become a basic expectation, and will keep being surprised that "adding something exciting" fails to lift its scores. The practical rule that follows from Kano is to re-classify a handful of key attributes each cycle rather than assume last year's map still holds, and to watch especially for yesterday's innovations that have slipped below the waterline of notice.

Related resources

• The SERVQUAL Gap Model, HEdPERF, and Service-Quality Thinking
• Expectancy-Disconfirmation and Course Evaluations
• Should You Use Net Promoter Score for Courses?
• Best-Worst Scaling (MaxDiff) for Course-Evaluation Priorities
• Why "Assessment and Feedback" Consistently Scores Lowest
• The Student-as-Consumer Effect on Course Evaluations

References

• Kano, N., Seraku, N., Takahashi, F., & Tsuji, S. (1984). Attractive quality and must-be quality. Journal of the Japanese Society for Quality Control, 14(2), 39–48. https://doi.org/10.20684/quality.14.2_147
• Herzberg, F., Mausner, B., & Snyderman, B. B. (1959). The motivation to work (2nd ed.). John Wiley & Sons.
• Madzík, P., Budaj, P., Mikuláš, D., & Zimon, D. (2019). Application of the Kano model for a better understanding of customer requirements in higher education — A pilot study. Administrative Sciences, 9(1), 11. https://doi.org/10.3390/admsci9010011
• Matzler, K., & Hinterhuber, H. H. (1998). How to make product development projects more successful by integrating Kano's model of customer satisfaction into quality function deployment. Technovation, 18(1), 25–38. https://doi.org/10.1016/S0166-4972(97)00072-2