The article says 'Changing the color of the "Proceed to checkout" button will increase purchases.' is a bad hypothesis because it is underspecified.

But what else is there to measure other than checkout button click count(and follow up purchases) to measure the effect of button color change?

Or perhaps this is not a robust example to illustrates undespeficaition?