What
"The percentage of decision outcomes that have been exercised by a test case suite."
BS 7925-1.British Computer Society Specialist Interest Group in Software Testing (BCS SIGIST)

A related metric is to measure Statement Coverage

Why?
A metric used to track the amount of testing conducted on one or more components, using branch testing. As this type of testing seeks, ideally, to execute every branch or decision outcome in the code, it is an important guide as to the amount of testing done, and the amount left. Especially that of Structural or White Box testing.

Who?
By personnel as close as possible to the code or testing. Thus the Test Analyst or tester. The coverage figure for individual aggregrates can be aggregated up of course to provide an overall management figure. However, the larger and complex a system is, it becomes harder to aggregrate.

Where?
At the devloping site.

When?
Ongoing throughout the lifecycle as components are identified, test analysis can begin, and the metric calculated.

How?
Two measurements are essential, the total number of decision outcomes, and the number to be tested, by the test suite. Note this does not have to be 100%. In non-safety critical systems, it will not be cost effective to test the last nth decision outcome.

At the start of testing the test analyst should know the percentage that will be tested. As testing progresses he can see the percentage that has actually been tested.