Conjointly offers various techniques for feature testing (including conjoint analysis, MaxDiff analysis, and Kano model). Of these two, MaxDiff with a single feature attribute and Kano are often seen as substitutes.

Kano has excellent face validity and is widely adopted in practice, especially in high-tech fields. But it has been criticised for the scale used and a bunch of other issues. MaxDiff, an even more popular method, on the other hand, has excellent theoretical foundations. So which should you use?

Empirical test set-up

We conducted a Kano model study with 100 U.S.-based managers, directors, and other decision-makers at companies who use teleconferencing software at least once a week. We tested 20 features. In our sample, users indicated they use the following software options:

The majority of them used more than one teleconferencing software programs:

We also ran a MaxDiff study on a convenience sample (also N=100). The MaxDiff question was focused on the importance of features:

Please review the following features of teleconferencing software and select the features that are most and least important to you.

The third study was also a MaxDiff test on a convenience sample (also N=100), but was focused on the trial-inducing potential of features:

Please review the following features of teleconferencing software and select the features you find most and least appealing to try.

Below are screenshots from the second and third studies:

Comparison of results

The table below compares the three studies. Features are sorted by the HB scores of the second test (MaxDiff focussed on the importance of features):

Below is a summary of Pearson correlations of the scores:

We see three interesting things:

1. Percentages of people for whom a feature is a must-have or performance are moderately correlated with the MaxDiff scores on the importance of the feature.
2. Percentages of people for whom a feature is attractive are modestly correlated with the MaxDiff scores on the trial-inducing potential of features.
3. Results from the two MaxDiff tests are not correlated.

Conclusions and recommendations

• Kano is a great first step for exploratory analysis of features. Because it does not use a unidimensional scale, you get multi-faceted feedback on features in a structured manner, without the need to run multiple studies using more specialised unidimensional scales.
• You can use MaxDiff as a more definitive follow-up test, but pay close attention to the MaxDiff question that you ask. What are you interested in? Getting new customers to try your product or keeping existing customers satisfied.

Please feel free to book a call with us if you want to talk more about Kano and MaxDiff pros and cons.