Download free Excel template for the Kano Model
Caption arithmetic (25 vs 200 is 1/8, not 1/7), table false-comparability (reliability coefficient vs replication accuracy are different metrics), unsourced "feature factory" claim (cut, no citable source in the evidence pack), UXDX stat missing limitation (flagged as industry-blog, not peer-reviewed), McFadden without IIA (added), "Google survey scientist" appeal-to-authority (named Chapman directly), long compound sentence in the alternative section (split), and the stray colon on the "Best for" row label (removed).
A product leader who searches for a free Kano Model Excel template wants a fast way to sort roadmap features into must-be, performance, and delight buckets before a planning meeting. Free templates exist today at Monday.com, Miro, ClickUp, Aha!, Parabol, and Conceptboard, and any of them will get you a filled-in classification of features in an afternoon. What none of them will give you is a measurement of how much any of those features would actually change what customers choose.
- A Kano Excel template classifies stated preference (how respondents predict they'd feel), not measured behavior (what they'd actually do).
- Published Kano category reliability sits at 0.61-0.73, below the 0.80 threshold considered good, per [Quant UX Blog's critical assessment](https://quantuxblog.com/critical-assessment-of-the-kano-model-part-2).
- Stable category assignment needs roughly 200 respondents; most teams run 20-30, per the same source.
- The method structurally cannot surface true delighters, since respondents can't rate a reaction to a surprise they haven't imagined.
- A randomized discrete choice experiment answers a different question: not "how would you feel," but "which action changes what you choose, and by how much."
## Where can you actually download a free Kano template?
Every major product tool ships one as lead-gen: Monday.com, Miro, ClickUp, Aha!, Parabol, and Conceptboard all publish a downloadable Kano matrix, usually a spreadsheet or a board template with the five categories (must-be, performance, attractive, indifferent, reverse) already laid out as a lookup table. The mechanic hasn't changed since Noriaki Kano introduced it in 1984: a paired functional and dysfunctional question per feature, tallied against a fixed scoring grid. Downloading one costs nothing and takes minutes. The scoring itself, and what the score means, is where the decision actually lives.
## What does a Kano survey measure, exactly?
A Kano survey measures a respondent's prediction of their own future satisfaction, not their revealed choice. Each question asks how someone would feel if a feature were present, and how they'd feel if it were absent, then maps the pair of answers to a category through a fixed lookup table. There is no effect size in that output and no confidence interval, because nothing was manipulated and nothing was chosen. The respondent is reporting an anticipated reaction to a feature they have never used, in a survey design that survey scientist Chris Chapman has formally critiqued for weak response-scale construction and item-wording confounds ([Chapman, Kano Analysis: A Critical Survey Science Review](https://research.google/pubs/kano-analysis-a-critical-survey-science-review/)). A completed template hands you a label. It does not hand you a measured causal effect on what someone would actually buy or use.
## How reliable is the category a Kano template assigns?
Not reliable enough to bet a roadmap on without more data than most teams collect. Published reliability coefficients for Kano category assignment run 0.61 to 0.73, below the 0.80 threshold survey researchers treat as acceptable, and stable, replicable categorization typically requires a sample around 200 respondents ([Quant UX Blog](https://quantuxblog.com/critical-assessment-of-the-kano-model-part-2)). Most teams don't get near that. A quick DIY Kano survey commonly runs on 20 to 30 respondents, a sample small enough that a feature's category can flip on re-run.
The instability hasn't gone unnoticed. A 2025 paper proposes an adaptive dual-response variant specifically to patch known category-assignment instability in the base method, evidence that a technique from 1984 is still being repaired ([ScienceDirect, 2025](https://www.sciencedirect.com/science/article/pii/S0969698925003753)).
## Why can't a Kano template find your true delighters?
Because delight, by Kano's own definition, is a reaction to something a respondent hasn't imagined, and you can't ask someone to rate their satisfaction with a surprise they've never conceived. The functional/dysfunctional question format requires the respondent to already hold the feature in mind before answering. That's fine for must-be and performance attributes, which people can reason about directly. It structurally excludes the category the method is most often used to sell: the delighter that nobody thought to ask for.
## What happens when you ship on a Kano-sorted roadmap?
Most shipped features still fail to deliver the value teams expected, with roughly an 8 to 10 percent success rate that UXDX cites as the norm for new product features, a figure from an industry blog post rather than a peer-reviewed study ([UXDX](https://uxdx.com/failure-rates/)). A Kano matrix doesn't change that base rate, because it never tested whether a feature would move a real decision. It tested whether respondents predicted they'd like it.
## What would a causal answer look like instead?
A randomized experiment, not a categorical survey. Instead of asking respondents to predict a feeling, a discrete choice experiment puts a randomized set of product configurations in front of respondents. It records which configuration they actually pick. Discrete choice models, such as McFadden's discrete choice estimator, Mixed Logit, or ICLV, then estimate the effect of each attribute on that choice. McFadden's estimator assumes independence of irrelevant alternatives (IIA): the relative preference between any two configurations is assumed unaffected by which other configurations are in the choice set. Mixed Logit and ICLV relax that assumption to allow for correlated preferences. Causal identification comes from the randomized manipulation in the design, not from the estimator itself. The output is an effect size with a confidence interval, and that interval covers the effect within the simulated population tested, not an unconditional bound on the real market.
Subconscious runs this kind of experiment on a simulated population and checks it against real human studies: simulated results reproduce the direction and outcome of the original human study 93 percent of the time on a validation set ([go.subconscious.ai/paper](https://go.subconscious.ai/paper)). That number describes replication against a held-out set of published studies, not a guarantee for a new, unseen market, and it comes with an honest limitation: some of those published studies could theoretically overlap with a model's training data, which is exactly what the replication protocol is designed to test against rather than assume away. Study-by-study replication results are public on the [leaderboard](/leaderboard).
| | Free Kano Excel Template | Randomized Discrete Choice Experiment |
|---|---|---|
| What it measures | Stated prediction of future satisfaction | Which choice respondents actually make when options are randomized |
| Output | A category label (must-be, performance, attractive, indifferent, reverse) | An effect size with a confidence interval |
| Reliability | 0.61-0.73 category reliability, a test-retest coefficient not directly comparable to replication accuracy | 93 percent replication accuracy against held-out human studies, a different metric than category reliability |
| Sample size needed | ~200 for stable categories; most teams run 20-30 | Set by experimental design, not survey convention |
| Detects true delighters | No, structurally excluded | Can test unstated or novel configurations directly |
| Cost to run | Free download, minutes to deploy | Requires a designed experiment |
| Best for | A fast, low-stakes gut check on a small feature list | A senior buyer who has to defend a roadmap decision with a measured effect |
## Should you still download the free template?
For a rough gut check on a handful of low-stakes features, yes. It costs nothing and it will surface obvious must-be attributes fast. What it won't do is give you a defensible answer for a decision that has real budget or roadmap weight behind it, because a lookup-table category isn't a measured causal effect and was never designed to be one. If the decision in front of you is which feature to actually build, a Kano matrix answers a different question than the one you're asking. Compare methods further on [methods and validation](/blog/methods-and-validation).
If you want to see where the causal approach stands on its own evidence, start with the [leaderboard](/leaderboard), which publishes study-by-study replication results rather than a single averaged number. For a walk through what a randomized experiment on your specific decision would look like, [meet](/meet) with the team.