Testing Conversion Variables

[Reactive]

What does everyone feel is the minimum number of visitors needed to accurately test a conversion variable?

For instance when testing different price points, would 2000, 10,000, or 50,000 visitors be a good minimum?

[crackity jones]

That depends on your conversion rate.

I have found that any sample size less than 1,000 is statistically insignificant. The problem is that the 1,000 refers to orders. This is something that can be tested at my 'day-job' - unfortunately my site has yet to generate a 1,000 orders in total. Jeff Bezos, I am not.

[Reactive]

Well my conversion rate is around .9% and I receive between 1,500 and 2,300 visitors per day.

I'm just wondering if I should wait 3 days, a week, or a month before I finishing testing a variable.