One strategy I came across in my interviews is email polling. This involves sending say 1000 emails/flyers A/B testing different prices to a list of potential customers and measuring the acceptance rates.
There are two versions of this test:
- Ask people which price they will buy at.
- Test which price people will buy at.
… then select the price at which profit is highest.
One benefit of these approaches is that there is a control experiment if the poll is done in a truly randomized fashion. That is there are no external effects that affects one set but not the other.
One problem unique to the first approach is that the price that people say they will buy at is different from the price at which they actually will.
This is a typical A/B test on pricing. There are several issues with this technique:
- The population that buys at the website is different from those that respond to the survey. The survey is sent to a population that was selected for different reasons from those that eventually purchase the device. What is worse is that it is usually sent to a list of people who showed an early interest for some reason. This is commonly known as sampling bias. The population the price is based on is different from the people who would actually purchase the product causing a loss.
- The survey is usually sent out once and the price is set. The optimal price is likely to change during the lifecycle of the product.
Early adopters behave differently from the main market. As the product gets accepted, it may become more obvious that it is valuable and people may be willing to pay more.
It is also possible that the manufacturer may discover that he can reach a sufficiently huge population at a lower price as the product becomes more mainstream. Perhaps they can become higher volume producers.