Last year a CEO said to me, "I would really, really like to understand the individual performance of people on agile teams."
This is a question that is anathema in the agile community. It's all about the team.
Mighty thinkers, like the Poppendiecks, have struggled mightily with this question.
It turns out that there is a simple, but gross, way to answer this question. As an added bonus you get to revisit your college algebra.
Recall that a system of linear equations can be solved using techniques that you learned in high school algebra:
(1) x + y = 100
(2) x = 30
From these two equations, we know that y=70.
How does this solve the individual performance question? Imagine that there is a Scrum team with two people, x and y. In the first Sprint, they work together to produce 100 units of value. This is represented by equation 1. In the second Sprint, person y is on vacation and person x alone produces 30 units of value. This is represented by equation 2.
This concept can be applied to any team that is organized in any way. It does not only apply to Scrum teams. The only requirement is that there must be a quantitative measure of value. The measure can be velocity or dollars or a KPI or something derived from an OKR. You get to choose. It could even be how happy the team is feeling on a scale of 1 to 50.
When you do this in practice you will find that the system of equations will be overdetermined. This simply means that there are more equations (Sprints or periods of time over which the data is collected) than there are unknowns (people). The standard approach, which you learned in college algebra, is to use the least squares method to find the "best" solution. Think of least squares as an objective function over the set of solutions.
Here is an example:
(1) x + y + z = 100
(2) x + y + 0.9z = 80
(3) x + y = 75
(4) x + y + z = 110
Here is how to read these equations:
(1) In the first Sprint, the three-person team produced 100 units of value.
(2) In the second Sprint, person z was out 10% of the time and the three-person team produced 80 units of value.
(3) In the third Sprint, person z was out the entire Sprint and the team produced 75 units of value.
(4) In the fourth Sprint, the three-person team produced 110 units of value.
The GRG Nonlinear solving method in Excel Solvr produces the following solution to this system of equations when minimizing the squared error:
x = 36.7
y = 36.7
z = 24.6
This solution can be read qualitatively as follows:
(1) Person x and person y are equally valuable on this team.
(2) Person z produces approximately one third less value than person x (and y) on this team.
I have never shared this approach with a client but I have used it to inform my coaching. The usual result is shocking: Not only are there people who are many times more valuable on a particular team than other people, but I often find that some people are contributing negative value!
This insight alone is enough to drive increases in agility at Fortune 500 companies over many eons.
 You may know someone who objects to using linear equations to capture team performance. They may suggest that equations of the form x+y+xy are a better model. The concept described in this article still applies but the math is harder and you will need to use a more capable solver. For ultimate expressive power apply machine learning techniques. Or the person may object to using math to capture team performance. That is an opportunity to be curious.