I first encountered the Kano model when working on improving the customer’s check-in experience at Copenhagen Airport. The model was conceived by Noriaki Kano in the 80s, and helps you analyze the customer experience of your product (or service), which ultimately allows you to invest more wisely in customer experience improvements.
The Kano model assumes three different attribute types – basic, performance, and delight – that collectively constitute the customer experience of your product.
The Kano model can help you analyze the customer experience of your product.
The three attribute types are mapped in a coordinate system with “Customer Satisfaction” up the y-axis and “Degree of Achievement” (how well a given feature is executed in your product) along the x-axis.
Let’s take an in-depth look at each of these attribute types along with some examples.
Basic attributes are typically taken for granted so customers only take notice when you fail to deliver.
Basic attributes represent features that are so basic to the product that your customers just expect them to work. These features are often taken for granted so customers rarely consciously look for them.
Examples of basic attributes are:
When dealing with basic attributes there’s not a direct relationship between the degree of achievement and customers satisfaction. When basic attributes are achieved your customer won’t be particularly satisfied, as he assumes it goes without saying. But when you leave out a basic attribute it doesn’t matter how well you otherwise perform or delight, the entire customer experience is broken. In this sense, it is difficult to actively use basic attributes as a competitive advantage, but if you fail it will put your company at a severe competitive disadvantage.
Performance attributes are rife with competition and it can be cost-prohibitive to be best-in-class on everything.
Performance attributes are features where there is a direct correlation between the degree of achievement and customer satisfaction. As a consequence companies tend to compete on these attributes, differentiating their product by spending more (or less) than their competitors on certain performance attributes.
Examples of performance attributes are:
The logic goes: the more legroom you offer your passengers, the more satisfied they will be. The steps are evolutionary, not revolutionary, and there’s typically a cost directly tied to the degree of achievement which is why performance attributes are natural candidates for competition (although winning that battle can be tricky).
Delight attributes are difficult to achieve but once you do the returns can be exceptional.
Delight attributes represent the unexpected – when you delight the customer by over-delivering or doing something out of the ordinary.
Examples of delight attributes:
Like the basic attributes, when dealing with delight attributes there’s not a linear relationship between customer satisfaction and the degree of achievement. When a delight attribute isn’t there the customer experience isn’t affected negatively because – by definition – delight attributes are never expected by the customer. However, when a customer is faced with a delight attribute it completely takes them by surprise, often resulting in over-excitement with your product, making it an effective engine for word-of-mouth.
One of the most important aspects with the Kano model is how time affects the attributes. Customer satisfaction with a given feature will deteriorate over time as companies start compete on the feature and customers get accustomed to it.
People quickly get accustomed with new features and raise their expectations accordingly which means you must constantly invent new ways to delight and perform.
When Gmail was introduced and offered 1GB free e-mail accounts it was revolutionary and the customer satisfaction was monumental – clearly a delight attribute. Nowadays, most free e-mail accounts offer similar storage capacity, to the point where ordinary users never have to worry about it. It has clearly moved from delight to performance, and is even starting to become a basic attribute.
So what today is a delight attribute that customers get truly exited about, will eventually degrade into a performance attribute as the competition starts to follow up and compete on providing the same thing (or even improving upon it). Eventually a feature reaches a state where further improvements are so inconsequential that they only matter when you fail to deliver – at this point it has become a basic attribute.
Hint: in an entertaining interview Louis CK describe this dilemma of ever-increasing customer expectations quite well.
So how do you use the Kano model in to your current or upcoming web projects? Besides analyzing the customer experience of your product, the Kano model can be useful when reflecting on how to invest more wisely in improving your customer experience. The following takeaways are particularly useful during this process:
Join 25,000+ readers and get Baymard’s research articles by RSS feed or
Topics include user experience, web design, and e-commerce
Articles are always delivered ad-free and in their full length
1-click unsubscribe at any time
© 2021 Baymard Institute US: +1 (315) 216-7151 EU: +45 3696 9567 info@baymard.com
MattFebruary 7, 2012
That was a delight to read. Thanks.
visitorFebruary 7, 2012
You guys keep on pumping out good quality information (that you guys do…and do it well), it is going to become ‘basic attribute’ around here ;)
Another great read…thank you!
Jamie, Baymard InstituteFebruary 7, 2012
Haha, thanks for the kind words!
We’ll try our best to make sure that happens ;)
Ifraz MughalFebruary 7, 2012
Great reference, Kano model brought to life.
Jonas BertéusenFebruary 13, 2012
Truly helpful artical. What’s your thought on how to measure and follow up on these metrics?
Christian, Baymard InstituteFebruary 15, 2012
It really depends on your industry, customers and product/service you offer , as it comes downs to how you measure “customer satisfaction” and gather customer insights. As far as the Kano model goes it’s up for you to decide.
Some ideas that comes to mind:
Measure sales, traditional quantitative CS surveys, qualitative user research (IDEO’s Method Cards is a good resource), “expert” panels, calculate and map technology/feature trend lines (typically for “performance” attributes), etc.
David TravisFebruary 13, 2012
Interesting article, many thanks. Minor quibble: ““Customer Satisfaction” up the x-axis and “Degree of Achievement”… along the y-axis” should read ““Customer Satisfaction” up the y-axis and “Degree of Achievement” … along the x-axis.”
Jamie, Baymard InstituteFebruary 13, 2012
Fixed. Thanks for the heads up :)
Tim from IntuitionHQFebruary 22, 2012
Another great article. I didn’t know much about the Kano model, so this was very insightful. Can’t wait to read another one of your articles.
Thomas DejanovicAugust 12, 2012
A good summary of the model, thanks!
In our office we’ve been having debates as to whether feature X is a delighter or is a performer – we’ve noted that our customers are not (yet) responding to the feature, but that could be because it’s a performer at a very slight slope and will thus take a lot more execution, or it could be a delighter that’s on the cusp of triggering a great deal of delight.
Any tips on what process to use to help classify my product features as being performers, delighters, or basic?
Christian, Baymard InstituteAugust 14, 2012
Hi Thomas,
That’s a great question. But unfortunately without a simple answer (besides the obvious “wait and see”).
There can of course be two complete non-Kano related reasons why customers aren’t responding: they haven’t discovered it yet (if it’s a new feature), they don’t understand or see it value (it’s not communicated clearly enough).
Kano-related you’re already on to two possibilities: most noticeably a performer at a very slight slope that will require a lot more execution before being measurable in practice.
A typical way of determine performance attributes are by the use of customer surveys to uncover the features that have the characteristic linear relationship between cost and degree of execution.
ShrikantJanuary 2, 2013
Thank you.
Kristin LudlowAugust 29, 2017
The Louis CK link no longer works – but you can pick it up here. Definitely worth a listen!
https://youtu.be/q8LaT5Iiwo4