Top questions & answers from this webinar
Q: There seemed to be a fifth question in the Van Westendorp model, but traditionally, there are only four. So what was that fifth question that you ask and is that by design?
Anwser: Yes, sometimes by design, I tend to ask a fifth question: Would you buy it at this price? That often gives me better survey data. I usually tend to change the wording of the Van Westendorp question in my price modeling. I veer towards asking questions about what you think is too expensive or too cheap, both in B2B and B2C, because we are often better at predicting other people. Then I ask if people would buy at that low or expensive price, giving you better data about pricing and the scope to purchase the product.
Q: While measuring opportunity cost, I’ve used the Van Westendorp model in the past, but there didn’t seem to be much difference between the two products. Do you think it would have been better to have used Gabor Granger price modeling in this case? Are there scenarios where you use both models or choose one versus the other?
Anwser: This is a great question. When the opportunity cost is almost the same, or there’s no negligible difference between the two's responses, one of three things is happening. Firstly, maybe the test is not sensitive enough. Secondly, people probably don’t see a value difference between the two. Or thirdly, when you have asked the question, you have not explained the value adequately. So this could be a communication error. In this case, you can ask the Gabor Granger question asking what you would pay for product A and what you would pay for product B. In this case, I would look at this data, and when people say they would pay similar for either product, I would ask a small subset of them the question ‘why’ to get to the bottom of this and see if the answer lies in the value of the product.
Q: You had mentioned sample sizes for sequential coming close to 150 to 400, and some of the other methodologies closer to 200. I have two questions regarding the N size: A) What drives those numbers and the differences across methods; B) At what point does the additional N size have diminishing returns - is ‘more’ always better?
Anwser: Someone already commented about the difference between B2B and B2C. In most cases, there isn’t a difference - it’s just people responding. But the most significant difference for me is the sample size. If you’re doing B2C, get a relatively higher sample size, and that’s not even expensive. In B2B, you pay higher amounts for the sample size, which sometimes drives what type of pricing study you conduct. There’s only one data point you’re getting from people in a simple Gabor Granger, so you will need a bigger sample size. If you’re using a little more complicated study where you use progressive price ranges, it’s OK to use a smaller sample size. This all is still simple yes or no answers. But if you also get degrees of thinking like you get with your conjoint analysis, you’re getting much better insights that can further drop your sample size. Depending on individual data points to a range of values, that’s what best defines your sample size. It all depends on your study and how you set it up.
Q: In BPTO - what determines how prices are changed?
Anwser: This is pretty intuitive. The one that is picked goes up in price, and all the others go down in price. The pricing mechanism has a step where a product has never been picked, there’s an amount changed. But from then on, it’s halving of the gaps between two values. So you can ask the question six, seven, or even ten times. You should know this though, the more questions you ask in pricing research, the less accurate it gets for the participants.
Q: Leading from that last line, then, how do you gauge what’s too much or too less, and how do you get optimum results in your pricing research?
Anwser: You want to make it as accurate as possible for the respondents. You want to replicate scenarios with what people are comfortable with. You want to replicate a real-life purchase situation. If respondents look at it as an in-depth situation, that situation brings out in-depth and near-real insights. You want to mirror a real-life decision.
Q: What pricing technique would you use in a B2B context where there’s not multiple buyers?
Anwser: What I typically suggest is A/B testing in these scenarios, but we don’t see that very much. So what I would recommend is conjoint because you are showing people different configurations. This is important because you want to know the price and understand different trade-offs in configurations. A lot of times, price isn’t the only vector in research. It’s important but not entirely top of mind. Since you are talking to fewer people - it’s essential to collect as many vectors as possible. It would be best if you met the need and in such cases, the price isn’t always the deciding factor.
Q: Which approach would you recommend in a case where the product is not ‘new news’, but is a premium product that has a very inexpensive substitute (e.g., a fancy coffee machine when Mr. Coffee is in the market)
Anwser: If I am launching that premium product, then I would have to say Van Westendorp. Because I am trying to test the communication of the said product. If it’s an existing product that is known, then if I am happy with the conjoint modeling, I can test the features and importance. But the sample is essential, and I will screen the people that are not willing to buy in this area. I will add those people back when I come to my model.