今天当Medium看到同样篇用研方法的牵线—— 卡诺模型（The Kano
Hygiene Factor in Quality)一平和。
Desired Features 期望功能
Required Features 必备功能
Delightful Features 魅力功能
Indifferent Features 无差异功能
Anti-feature Features 反为效应
Satisfaction Coefficient 与 未如意系数 Dissatisfaction
以下是原文内容 Let’s Go：
The Kano Analysis
A Better Way Discover What Users Really Want From Your Product
You’re on the design team for Crunchrr, a new app that helps users
discover cereals they’ll love. Users can:
– Create a profile and connect with others
– Discover cereals based on their preferences
– Rate and review cereals
Crunchrr is in the hands of some early adopters who are loving its core
features. Things are going great. That is, until the requests start
Annelise from marketing says: “Crunchrr needs a map view so users can
see where each cereal is made. People are really interested in where
their food comes from nowadays, so this is really a must! Besides, every
app has a map view.” Kevin from sales was at a meeting with a potential
advertiser who asks: “Where’s the chatbot? You can’tnothave a chatbot.
Conversational UI is the future!”
One of your early adopters pings you to suggest: “There should be a
button so I can email the cereal maker to request a gluten-free
version.” Another one says: “Maybe there could be something like Shazam
for cereal. That way, if I’m in a restaurant I can take a picture of
what the person at the next table is eating and it’ll show me what that
The next thing you know, your backlog is a gaggle of suggestions,
requests, and demands. It seems that everyone has brilliant idea that
justhasto go into the next release.
This can’t be avoided. Everyone has an opinion and given the
opportunity, they’ll express it. And people easily fall into a “more is
better” mentality. More features equals a better product, and the more
of each feature, the better.
The obvious problem is that you can’t deliver on every request. Not only
that, but all ideas aren’t created equal, and users are often at a loss
as to how to articulate what they really want and need. On the other
hand, internal stakeholders tend to view features in the narrow context
of their own interests. How do you stop the madness?
“The most important thing that a team can do to help their design is
to say no to almost any idea for a feature”
— Jared Spool
You need a way to predict user satisfaction that lets you prioritize
feature releases and even re-evaluate existing features. And you need
hard data to support your decisions about what goes into Crunchrr and
when. That’s where theKano
The Kano Model
In 1984 professor Noriaki Kano presented a model that predicts how
satisfied people will be with a product based on its features. Since
then, the Kano Model has become a standard design tool because of how
effectively it can make typically invisible ideas about quality visible.
The core principle of the model is that satisfaction can be plotted
along five distinct
Curve 1: Desired Features
Remember when I said more isn’t always better? Well,sometimesit is it
is. More storage space or battery life is better. Faster download
speeds? Better. These are all examples of where the user will usually
express greater satisfaction in direct proportion to how much of the
feature they get.
With desired features, satisfaction is directly proportional to feature
In the case of Crunchrr, desired features could be:
– Speed and responsiveness
– Number of users to connect with
– Suggestions based on stated preferences and past browsing behavior
– Options for quickly zeroing in on a kind of cereal (sorting,
– Size of cereal selection
Curve 2: Required Features
Required features are the ones users expect and take for granted.
With required features satisfaction levels off once the basic need has
Users are dissatisfied when a required feature is not present and
satisfied when it is. But that satisfaction levels off after a certain
point. This makes sense when you think about it. If a wheel doesn’t
roll, it will cause dissatisfaction. If it does roll, it will cause
satisfaction. But it’s hard to get anyone excited about a wheel that
rollsreally, reallywell. In the case of Crunchrr, as with most other
apps, this could mean things like:
– Reliable uptime
– Ability to create a profile
– Easy log in/out
Curve 3: Delightful Features
Delightful features are the ones that make an app fun to use and give it
a personality. They’re the features you love, but don’t expect. It could
be as simple as when the login form appears to shake its head when you
enter the wrong credentials. Or it could be the tone of the writing or a
fun mascot character or some unique interaction.
Users are satisfied with delightful features, but are not dissatisfied
when they are absent
As you can see from the graph, users express increased satisfaction with
delightful features. But there’s no dissatisfaction when they’re not
present. Also, as with required features, there’s a limit to just how
delighted a user can be. After a certain point, there are diminishing
Annelise’s map view is probably an example of a delighter because it’s
little more than eye candy, and it certainly isn’t solving any of the
currently defined business needs for Crunchrr.
Delightful features are an important part of the user experience, and
shouldn’t be ignored. Butthey come with a shelf
in part because they’re so easily imitated. For a while, the swiping
interaction was a big part of Tinder’s unique identity. Now, Tinder is
justone of many
users can swipe left or right. In other words, over time, delightful
features go on to become desired or even required features.
Curve 4: Indifferent Features
These are features the user simply doesn’t care about either way.
Whether they’re implemented fully or not at all, they won’t change
users’ opinions about the app, or change how they use it.
Neutral features don’t affect satisfaction one way or another
Curve 5: Anti-features
Anti-features are the features that users actively dislike.
And the more these features are implemented, the greater the
dissatisfaction. Anti-features are like the mirror opposite of desired
Anti-features are the ones that frustrate or annoy users.
Dissatisfaction is directly proportional to implementation
Putting it All Together
Looking at all of these features together not only provides a clear
pictorial representation of how features will be perceived, but also
helps you figure out strategic direction.
The complete Kano Model diagram
Desired Features:Resources should be invested heavily in these features,
because they are key to user adoption and retention, as well as
Required Features:Resources should be invested heavily in these
features, but only until basic needs have been met.
Delightful Features:It’s fine to invest resources here, but not at the
expense of desired and required features. However, delightful features
are often key differentiators that can build loyalty and buzz.
Indifferent Features and Anti-features:Resources should be invested only
in identifying these so as not to waste cycles on building and
By now I hope you’re sold on the Kano Model. Then the next question is:
How do you find out which features belong to each category? That’s where
the Kano Analysis comes in.
The Kano Analysis
To find out which features belong where, we need to ask our users. But
remember, users are not usually great at identifying or expressing what
they really want and need. The Kano Analysis accounts for this by asking
questions in pairs: afunctional questionfollowed by adysfunctional
question. Let’s go back to Annelise’s suggestion of a map view for
Crunchrr. We could ask a question pair about this feature like this:
If Crunchrr let you see on a map where a brand of cereal is made, how
would you feel?
If Crunchrr did not let you see on a map where a brand of cereal is
made, how would you feel?
For both functional and dysfunctional questions, users must choose one
of the following answers:
– I like it that way
– I expect it that way
– I am neutral about it
– I can live with it that way
– I dislike it that way
You would prepare an entire questionnaire in this style for each of the
features in your backlog. Each user’s answers can then be analyzed by
plotting its outcome in the following table.
The analysis table tells you where a user would place a feature in the
Kano Model based on how the functional and dysfunctional responses
It should be clear that if a user likes it when the feature is present
and dislikes it when it’s not, then that is a desired feature. The
designation ofquestionablehappens when the answers apparently contradict
each other. (This is often the result of the user not understanding the
Great. We’re almost done. The final piece is to aggregate all of the
survey responses to find the overall results for each feature.
(Alternatively, you could break this down even further and aggregate
responses based on personas.)
After you’ve aggregated all of the responses, you’ll calculate the
satisfaction and dissatisfaction coefficients. The satisfaction
coefficient is a number between 0 and 1: the closer to 1, the stronger
the influence on satisfaction. The dissatisfaction coefficient is a
number between 0 and -1: the closer the closer to -1, the stronger the
influence on dissatisfaction. We calculate the coefficients with these
Let’s say that the aggregated responses for the map view breaks down
That would give you these results:
Satisfaction: (4 + 5) / (4 + 5 + 12 + 23) =0.2045
Dissatisfaction: (5 + 12) / (4 + 5 + 12 + 23) * (-1) = -0.3864
As you can see, the map view feature is having a significantly stronger
influence on dissatisfaction than on satisfaction. This clearly
indicates that we should leave it out of Crunchrr. Sorry, Annelise!
(Actually, if you saw these results in the real world, you wouldn’t even
need to calculate the coefficients. Seeing 31% anti-feature and 25%
questionable is enough to tell you not to include this feature. I used
these exaggerated figures to highlight the differences produced in the
Other times, the coefficients will show little difference in influence.
Cases like those will require a judgement call or re-testing.
A Kano Analysis is cheap and easy to perform and provides clear vision
into what users actually want and expect from your product. It also
provides hard data, which breaks everyone out of the trap of biased or
shortsighted thinking. There’s no need to argue and debate with internal
stakeholders about which features are in or out. The numbers don’t lie!
Brian O’Neill @brianeoneill is a designer in the San Francisco Bay Area,
currently at NVIDIA.
curves go by many different names, depending on the source. I picked
these names arbitrarily. In the end, it doesn’t matter what they’re