Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

How to decide whether or not to implement condictional pricing in CBC?

Dear Sawtooth Community,

I am implementing a CBC in which I have different package sizes and brand levels.
In order to keep analysis simpler I am considering to not implement conditional pricing (cp). However, I am wondering if NOT implementing cp in the described case poses a risk to the data quality (e.g. in the case that respondents would always choose the unrealistic concepts)?

Further, I have read that it is possible to make the price conditional on up to three attributes (in my case it would be two). However, I have also read that it is best for further analysis to not make the price conditional on more than one attribute. What is your experience?  

Thank you so much in advance!
Best regards
asked Dec 13, 2019 by Mel (250 points)

1 Answer

0 votes
Best answer

Different package sizes is the most common place we see conditional pricing used, and it might well not seem realistic for respondents to see individual products priced at multi-pack prices and vice versa.

The reason we advise price and only one conditional attribute is that IF the main effects don't do a good job of accounting for utilities, then you're going to want to estimate an interaction between the conditional attribute and your price attribute.  Two-way interactions are easily done in our software, but if you go up to a three-way or a four-way interaction because you included 2-3 other attributes in determining your conditional prices, then, you would have to export your design matrix, add all the interaction terms manually and then run the analysis in our stand-alone CBC/HB software.  Trust me, you do not want to code up your own three-way and four-way interactions if you can avoid it!
answered Dec 13, 2019 by Keith Chrzan Platinum Sawtooth Software, Inc. (110,575 points)
selected Dec 13, 2019 by Mel
Thank you Keith!

You mentioned, that without cp some concepts might not seem realistic for respondents. However, in my case the product is relatively new and respondents would not necessary think, that the concepts are unrealistc.

So do you see any further risk, if I decide to NOT implement cp - such as a risk to the data quality (e.g. in the case that respondents would always choose the unrealistic concepts)?