Kano Model and Faster Horses

Question

I should probably mention one thing before I start: When judging if a solution is "right", I rather believe in seeing people's actual behaviour than asking them about it.

Now let's move on to the question: I have used the Kano model a few times in the past and there was always one thing lingering in my head: We're basically asking people what they like instead of watching what works for them. Usually you don't get to a state where you can show people a prototype of a feature when running a Kano study, instead you often show them just a sketch or a mock up. I decided to run a little experiment lately and it made me question the Kano model as a whole.

I did build an interactive prototype with certain features in order to find out what works for people who use the product. In that prototype I had Feature A, Feature B and Feature C. All of the features were relatively new to them and it's unlikely that they have seen them in a product before. I gave them realistic tasks and watched how they used the prototype. Watching them, almost everybody used Feature A heavily and Feature B & C only a little bit or not at all.

After they finished the tasks, I asked them the questions from the Kano model and was shocked by the results:

1. Feature A turned out to be Indifferent.
2. Feature B was a Performance feature.
3. Feature C was Attractive.

I ran the study with eight people and the results were pretty clear, there wasn't much deviation. People basically said that they don't really need the feature that they used the most to finish a task. They also said that they would love the features that they barely used.

Did anyone run into that problem before? Did I miss something in my setup/observations?

Answer 1

Hey, good question. I'd argue how much of a 'problem' this is really depends on what your desired outcome is.

In the Kano model, satisfaction and functionality are on separate axis. For a utility product, I believe the model can under-value satisfaction because the scales are linear. If we step away from the solutions-finding models and look instead at the surveys in use for scoring customer loyalty, they are strongly weighted so that only the highest scores mean anything (6-10 range). For example the NPS scoring system. So if you are trying to create a product that is a boring utility thing (think spreadsheets), people will assume a whole bunch of functionality, and satisfaction scores won't ever rank particularly highly.

On the other hand, if your goal is to build an easily marketed solution, then perhaps the Kano model is telling you something useful. For years, Audi have sold new family cars on the [attractive] vision that you can drive into the mountains and go skiing. And not on the features people have come to expect, and so are indifferent about.

The danger with building to be marketed, is the decay of satisfaction. If we take a look at how Apple marketed the new iPad pro - again the marketing is all about [attractive] composing sick beats and creating beautiful art, and not the features people now expect from a tablet. So just like you identified - people are attracted to a product by shiny features they barely use.

I'd take the model with a pinch of salt, and keep in mind what you are trying to achieve.

Answer 2

The Kano model provides a way to think about how products features can be perceived by users, but like most models it is an idealistic view of how something might actually work in real life. I quite like the explanation of the model provided on The Complete Guide to the Kano Model where it goes into some detail about how to set up this type of study.

Long story short, there are some factors that make the analysis of results more complex than it first appears, and these are some assumptions that we normally make about users and product features.

For example, it is possible for users to have different views about the same feature if you are comparing a beginner to an expert user.

It is also important to remember that the model actually caters for the fact that people will change their perception of various features over time as well, such that a 'shiny' feature will eventually become standard or a must-have.

Also, remember that you need to validate the type of behaviour by adjusting the level of functionality and observing the impact of that change to be more confident that this is indeed what people feel about that particular feature. Testing and scoring the feature at a single data point won't necessarily tell you what the pattern or trend is.