Managers like metrics because it gives them a (seemingly) objective way to measure the performance of their people and reward/correct them. There are better ways to reward or correct performance.
As an exercise in choosing metrics for measuring your team members, ask yourself this: "How can this metric be gamed?" If someone can gain financially from gaming a metric, it will be gamed.
You should also ask yourself why you want to perform such a task in the first place. If you are doing it to decide who needs to get cut from your team or receive an amazing bonus or raise, there are better ways.
James Bach recommends managers walk around in the developer pit and talk to people, listen to people, and get a sense of how the team is functioning on a regular basis. This will tell a manager more about his people's performance than any stack-ranked arbitrary metric.
A metric that measures team performance (developers and testers as a single team) is a list of bugs reported by customers, sorted by release. If the issue was found pre-production and not addressed, it still counts. This shows the delta between what the customer wanted and what we thought they wanted.
Since top management likes to measure the money, I have in the past used a story-based metric to show top management how valuable my team was. At raise/bonus time I would poll the bug repository for all the urgent issues opened by each tester. I would hand the list to each tester and have them guesstimate the value of the top 10 or so of each of their found issues had they not been found. Value can include such things as lost revenue, lost productivity, defecting customers, loss of company reputation, or anything else that can be quantified in a dollar amount. The dollar amounts were added up by each tester and always exceeded their annual salary. The lists of bug IDs, titles and dollar amounts were handed up the management chain as a demonstration of test team members' effectiveness.
There were many estimations and assumptions that went into each line item; I would help them come up with reasonable assumptions. These were available to anyone who wanted to challenge the assumptions. Nobody ever did.
