The real problem for any manager, owner or executive when trying to quantify the return on investment from any social media investment is that the results are highly dependent on the assumptions. Change the assumptions and you change the "return" on any campaign or channel.
That can be especially tricky when trying to figure out the value of "avoided" activities. Any customer support group can estimate the cost of an average call simply by dividing wages or contract costs with the total number of "completed" calls. None of that will measure the effectiveness of call handling, though.
A firm might rationally want to process more calls per hour. What isn't so clear is how the value of those calls might change if the metric is simply "calls per hour." Handling calls faster might mean higher rates at which customers depart, fewer "account saves" or less "incremental new sales."
In the U.S., the average support call cost is approximately $10 to $25 per call, depending on the product, services and the vertical. So one way to create a metric is to forecast the number of support calls a firm believes it will get in the future, implement a social media program to circumvent those calls, and then measure the difference between the number of calls the firm actually received, versus the number it expected.
You see the conundrum. "Success" depends on the assumptions about future call rates. Set a high-enough expected future rate and a firm can "succeed" with or without a social media program. Set a low-enough rate and a firm will "fail," no matter how good its performance at avoiding calls in the first place.
It's a good thing, and often necessary, to measure social media effectiveness. The problem is simply that success or failure mostly is determined by the assumptions one makes.