Are ROAS estimates reasonable?

[maswiebe]

You estimate that TV has an ROAS of 11.04: a $1 increase in TV adspend increases sales by $11. In percentage terms this is a 1004% return, which is incredibly high; in the stock market, 10% is a good return.

Do you think this is reasonable? How do you correct for selection bias and reverse causation? What's your statistical power to detect smaller effects like 10% (or ROAS = 1.1)?