You still have the problem of distinguishing real changes in "happiness" with noise in your measurements, so the statistical techniques underlying A/B testing a

You still have the problem of distinguishing real changes in "happiness" with noise in your measurements, so the statistical techniques underlying A/B testing are still useful and relevant. The question I want answered is how well this metric will correlate with other metrics I care about (like Customer Lifetime Value). If it's not predictive of business success it's of no use.