Kano Model
The Kano model classifies product attributes by how customer satisfaction responds when the attribute is present or absent: a missing must-be attribute causes dissatisfaction, a one-dimensional attribute raises satisfaction in proportion, and an attractive attribute delights when present but is not missed when absent [1]. Noriaki Kano and colleagues published it in 1984. It answers a question that “satisfaction is performance against expectation” leaves open: why can delivering everything customers asked for still disappoint them?
1. Five categories
Kano and colleagues named three main categories and two minor ones [1]:
- A must-be attribute is taken for granted: its absence makes customers dissatisfied, and its presence adds nothing.
- A one-dimensional attribute works like a slider: more of it means more satisfaction.
- An attractive attribute delights when present and is not missed when absent.
- An indifferent attribute changes nothing either way.
- A reverse attribute makes some customers less satisfied when present.
Practitioner names on the figure: delighters = attractive, satisfiers = one-dimensional, basic needs = must-be. The curves show shape only, and the migration arrows are not in Kano’s or Sauerwein’s papers. [1], [2]
2. Classifying an attribute with two questions
Kano classified attributes from a pair of questions per attribute: a functional question (“how do you feel if the product has it?”) and a dysfunctional one (“how do you feel if it does not?”), each answered on the same five-point scale: I like it, I expect it, I am neutral, I can live with it, I dislike it [1]. The pair of answers gives the category through the evaluation table: find the row for the answer about presence and the column for the answer about absence [3].
| Present ↓ / Absent → | Like | Expect | Neutral | Live with | Dislike |
|---|---|---|---|---|---|
| Like | Q | A | A | A | O |
| Expect | R | I | I | I | M |
| Neutral | R | I | I | I | M |
| Live with | R | I | I | I | M |
| Dislike | R | R | R | R | Q |
A attractive, O one-dimensional, M must-be, I indifferent, R reverse, Q questionable: the two answers contradict each other, so the response is discarded.
For example, a team asks 50 users of a video-streaming app about “skip the intro”. Their answer pairs (present, absent) are: like and neutral, 18; like and live with, 7; like and dislike, 9; neutral and neutral, 11; expect and dislike, 3; like and like, 2. The table turns them into A = 18 + 7 = 25, O = 9, I = 11, M = 3 and Q = 2. The two questionable answers are dropped, and the feature takes its most frequent category: attractive, with 25 of 48 valid answers. The same survey gives “resume where I stopped” M = 38, O = 6, I = 4, A = 2, so must-be, and “4K picture” O = 31, A = 9, I = 6, M = 4, so one-dimensional. When two categories are close, Sauerwein and colleagues break the tie in the order must-be, one-dimensional, attractive, indifferent, because a failed must-be cannot be compensated [2]. Chapman and Callegaro report that category assignment may be unreliable with samples under 200 [4], so treat a 50-person survey as a first signal. Two satisfaction coefficients per feature read the counts as an investment case: better = (A + O) / (A + O + M + I), how much having it raises satisfaction, and worse = −(O + M) / (A + O + M + I), how much lacking it lowers satisfaction [2]:
| Feature | Better | Worse | Reading |
|---|---|---|---|
| Resume where I stopped | 8/50 = 0.16 | −44/50 = −0.88 | insurance: getting it right buys little, getting it wrong is punished hard |
| 4K picture | 40/50 = 0.80 | −35/50 = −0.70 | both directions matter |
| Skip the intro | 34/48 = 0.71 | −12/48 = −0.25 | upside with little downside |
The team concludes that resume must work before anything else, 4K is worth steady investment, and skip-intro is a cheap way to stand out.
When to use it — and when not
- Use Kano to prioritise features and to set quality targets: must-be attributes need a reliability threshold, such as “resume works in at least 99.9% of sessions”, and one-dimensional ones a target on a performance scale.
- Do not use one survey forever: categories drift, and yesterday’s attractive feature becomes tomorrow’s must-be, as phone cameras, free HTTPS and dark mode did (our examples).
Limitations
Categories hold for a group of customers at a time, not for everyone: a feature can be must-be for professionals and indifferent for casual users, so the most frequent answer can hide two segments. The model says which way satisfaction moves, not by how much.
How solid is this?
- Where it comes from. Kano and colleagues’ 1984 paper, in Japanese, introduced the categories and the question pair [1]; Sauerwein and colleagues’ 1996 ski study (over 1,500 customers) gives the coefficients and the tie-break rule, and reports, after Griffin and Hauser, that 20 to 30 interviews per homogeneous segment surface 90 to 95% of the requirements [2]. That figure is about finding requirements, not about classifying them reliably.
- What is contested. Scoring rules beyond “most frequent category” differ between authors, and Chapman and Callegaro question whether a method built for durable consumer goods fits technology products [4]; the drift from attractive to must-be is widely observed but rarely measured.
- What we do not hold. The original paper’s text. The evaluation table is the form popularised by Berger and colleagues [3], checked against several published reproductions; the streaming survey is constructed.
References
- N. Kano, N. Seraku, F. Takahashi, and S. Tsuji,“Attractive quality and must-be quality,” 1984.
- E. Sauerwein, F. Bailom, K. Matzler, and H. H. Hinterhuber,“The Kano Model: How to Delight Your Customers,” in Preprints Volume I of the IX International Working Seminar on Production Economics, Innsbruck/Igls, 1996, pp. 313–327.
- C. Berger et al.,“Kano’s Methods for Understanding Customer-Defined Quality,” Center for Quality of Management Journal, vol. 2, no. 4, pp. 3–36, 1993.
- C. Chapman and M. Callegaro,“Kano Analysis: A Critical Survey Science Review,” in Proceedings of the 2022 Sawtooth Software Conference, 2022.
Disclaimer: AI is used for text summarization, polishing and explaining. Authors have verified all facts and claims. In case of an error, feel free to file an issue.