In the previous post on PPC cost estimation I made a mistake.   Since twice as many people are likely to opt in as are likely to take the long survey (7 to 10 questions), I typically DIVIDE the cost per survey by two, not multiply it by two.

Thanks to everyone who pointed this out!