Many teams have multiple team members who split time between projects. In most cases, it is better to have fewer people full time on a project than more people part time. For example, instead of having eight people working part time, try having four people work full time.
Why? We lose 20 percent efficiency each time we multitask, according to Mike Cohn's book Succeeding with Agile. Count the average percent of time each person is allocated to the project. Averages less than 80 percent should be looked at to see if the team is being pulled in too many directions.
This is the ratio of work planned compared to potential capacity. In theory, people could allocate up to 54 task hours per two-week iteration, but the team only plans an average of 30 hours. Then there is a problem with planning.
Fill your plan until you have 70 to 80 percent of the team's time accounted for. Eighty hours for a two-week sprint is not ideal because other important work is not represented by tasks (such as mandatory training, email, unexpected maintenance work).
In addition, have the team own its total task hours and let them "pull" work when they are ready. Assigning work to individuals may create silos and is based on imprecise estimates.
Compare the number of task hours left at the end of the iteration to the total planned. There should usually not be any hours left, but if the team meets this goal every time, it may not be planning aggressively enough. It's healthy for teams to have a small number of hours, say 5 percent, left over on a small number of their iterations (10 percent or less).
This shows how much unplanned work was added during the iteration.
In my experience, it is okay to add up to 15 percent because planning is based on approximate estimates and technical execution of the project may reveal new subtasks. But if more than 15 percent has been added, it's a sign the team is either not planning enough or they never say no to new requirements. Either way, they lack the ability to plan with enough diligence or to pause new work until future sprints.
What do you think? Have you used these metrics? What metrics do you use to help spot trouble?