I don't think the intention is to run this test for "an extended period of time." You want to gather enough data for it to be significant and then move on to th

I don't think the intention is to run this test for "an extended period of time." You want to gather enough data for it to be significant and then move on to the next price you're testing. But, it's a very good question about starting high or low. The common wisdom is to always price high and then move lower. I've seen this work well for sales by being able to offer a discount to close a deal. There's also the perception that expensive = high quality. The only real way to know for sure, however, would be to duplicate the test and go high-to-low on one and low-to-high on the other.