Prioritization
Kano Model
Sort features by how they drive satisfaction: must-be, performance, delighter.
Interactive model
Kano Model, at a glance
Satisfaction = f(feature performance)
Separate table stakes, linear performance, and genuine delight.
How to read the model
What it is
The Kano Model is a feature-prioritization and customer-satisfaction framework developed in 1984 by Professor Noriaki Kano at the Tokyo University of Science (his paper "Attractive Quality and Must-Be Quality," and popularized for English-speaking product teams by Berger et al. in the 1993 Center for Quality Management Journal). Its core insight is that satisfaction is asymmetric: the presence and the absence of a feature move the satisfaction needle by different amounts, so not all features are worth the same investment. Kano sorts features into categories — Must-be (Basic), Performance (One-dimensional), Attractive (Delighter), Indifferent, and Reverse — based on how their presence AND absence each affect customers. It matters because it explains why teams can ship polished features and still have unhappy users: they invest in delight while a broken basic quietly tanks satisfaction. It forces you to separate "table stakes" from "differentiators."
When to use it
- Use it for feature-prioritization and product-design interview prompts ("How would you prioritize the roadmap for X?" / "What would you build next for Product Y?"), and whenever you need to justify sequencing by customer impact rather than gut feel — especially to argue that a flashy feature should wait until a basic expectation is solid.
When not to use: Skip it when priorities are dominated by effort, reach, or business ROI rather than satisfaction (use RICE or opportunity scoring), or for a v1 MVP where the categories are obvious and a full survey is overkill.
The steps
- 1List candidate featuresEnumerate 10-25 concrete, mutually-understandable features (e.g. 'offline playback', not 'better UX') so each can be evaluated in isolation without overlap.
- 2Ask the functional AND dysfunctional question per featureFor every feature ask a matched pair — 'How do you feel if the product HAS this?' and 'How do you feel if it does NOT?' — each answered on the same 5-point scale: I like it / I expect it / I'm neutral / I can tolerate it / I dislike it.
- 3Classify each feature via the Kano evaluation tableCross-tabulate the two answers on the standard grid: Like+Dislike = Performance; Expect+Dislike = Must-be; Like+Neutral = Attractive; Neutral+Neutral = Indifferent; Dislike+Like = Reverse (they want the opposite); Like+Like or other self-contradictory pairs = Questionable. Assign the modal category across respondents.
- 4Quantify with the Better/Worse (CS) coefficientsCompute Better = (A+O)/(A+O+M+I) and Worse = -(O+M)/(A+O+M+I), where A/O/M/I are the counts of Attractive/One-dimensional/Must-be/Indifferent responses. 'Better' (0 to 1) is the upside if present; 'Worse' (0 to -1) is the downside if absent — plot them to see each feature's true leverage.
- 5Sequence the roadmap by categoryGuarantee every Must-be first (they are satisfaction floors, not selling points), fund Performance features to competitive parity or better, sprinkle in a few Attractive delighters to differentiate, and drop Indifferent/Reverse items.
- 6Re-run periodically because categories decayDelighters commoditize into expectations over time ('wows become wants become musts'), so treat classifications as a snapshot and refresh them as the market shifts.
Worked examples
Netflix runs a Kano survey on five features. 'Video that plays without buffering' scores Better +0.35 / Worse -0.90 — a classic Must-be: nobody praises smooth playback, but buffering triggers cancellations, so it's a non-negotiable floor, not a roadmap headline. 'Personalized recommendations' comes back Performance (Better +0.75 / Worse -0.55) — more accuracy linearly lifts satisfaction and retention, so it deserves continuous investment. 'Downloads for offline viewing' lands Attractive (Better +0.70 / Worse -0.15): its absence upsets few, but it delights travelers and differentiates against rivals. The insight: pouring engineers into recommendations while playback stutters would be a category error — fix the Must-be first, then compete on the Performance and Attractive features.
Spotify weighs 'AI DJ / smart mixes' against 'gapless crossfade between tracks.' The survey classifies AI DJ as Attractive (Better +0.68, Worse -0.10) — a delighter that generates buzz and word-of-mouth but whose absence nobody complains about — and crossfade as Indifferent for most users (Better +0.20, Worse -0.15). The Better/Worse plot surfaces the real trade-off: crossfade is polish almost no one will notice, while AI DJ is a low-risk, high-upside differentiator. The seniority signal is naming that AI DJ will decay — today's delighter becomes tomorrow's baseline once competitors copy it — so Spotify must keep a delighter pipeline rather than treating it as a one-time win.
Common mistakes
- Treating Kano as a single-question importance survey — the entire model depends on the MATCHED functional + dysfunctional pair; ask only 'how important is this?' and you can't detect the asymmetry that makes it Kano.
- Confusing Must-be with 'high priority.' Must-be features max out satisfaction at 'not angry' — over-investing in them past the expected bar yields zero upside, which is exactly what the model warns against.
- Chasing delighters while a basic is broken. Attractive features can't compensate for a missing Must-be; a magical feature on top of a buggy core still nets a dissatisfied user.
- Treating classifications as permanent. Delighters commoditize into expectations (two-day shipping, dark mode, auto-suggest all made that journey), so a stale Kano chart quietly misleads the roadmap.
Interviewer tip
Naming the asymmetry out loud — 'this is a Must-be, so its ceiling is neutral, not delight; I'd fund it to parity and no further' — signals you think in satisfaction curves, not flat feature lists. Bonus seniority points for noting that today's delighter decays into tomorrow's baseline, which shows you reason about a category over time, not just at one snapshot.