Can We Simplify Data Collection to Understand Customer Preference?
The canteen today looks very crowded. Various menus were provided in the canteen with the fractional cost compared to if the employee buy the items in the restaurant. Ran select one menus. She almost couldn’t find any empty chair. Luckily, Dess and few of his team already book a table. They are able to sit together.
“How was your weekend Ran?” Dess ask Ran when she is about to enjoy her lunch.
“Wonderful. I am able to play table tennis with my kid and also visiting book store”.
“Wow… productive weekend. How about you Ray?”
Ray slurping his noodle and great enjoyment, “You know Dess, I am still single and many of my friends already starting a family. So, I have the time to think about work”.
Dess chuckle to hear Ray grumble, ”Interesting, what are you thinking about?”
“About our last discussion, conjoint analysis. Remember how we collect the data? We have levels and attribute of the product that we want to research. When the levels and attributes are too many, we create orthogonal survey, so we can minimize the number of survey questions to the customers to rank their preference.”
“Yes you are right,” respond Dess .
“However,” Ray continuing his chain of thought, ”often times, the number of combinations left for the customer to rank is still to many. And I understand that ranking preference is not very easy when the combination of attributes not very different.”
“So, I was thinking, is there a way to collect the data from customers where they have more convenience to select a products with its attributes?”
“Very good observation Ray, as expected from you.” Dess then continue, ”On our previous example, we use full profile conjoint analysis where we use a customer’s ranking of product profile to determine the relative importance of an attributes for the customers. And you are right, when there are too many choices, the result might become unmanageable.”
“ You see, there are various way to collect customer preference data. Other thank ranking the product profile, we could ask the customer to compare between two product and ask to select one. This method called Pairwise Comparison. But this method is limited to one attributes for each selection. Another method is asking the respondent their preferred product profiles amongst profiles. Here is the example.”
“And,” Dess continue, “we don’t left the respondent to only answer one questions. The respondent need to answer next questions with different product profile and choose one from this options. And then another similar questions with different set of product profile.”
“Once we collect enough data, we could then estimate the relative importance the customers attaches to product attributes and how the customers rank the levels of each product attributes.”
Ray looks very interested. He very keen to know this new method and how he can calculate the importance of product attributes and how we can draw conclusion from the result.
“It looks very interesting Dess, should we comeback to our drawing board?”
“Sure, lets do this in the evening after we complete our call with the client.”
Source:
Winston, W. L. (2014, January 8). Marketing Analytics: Data-Driven Techniques with Microsoft Excel.
https://www.opinionx.co/research-method-guides/discrete-choice-modeling. Accessed 5 May 2024.
