Any recommended template for capacity calculation?

I think mostly trying to micro-manage capacity isn’t worth it.

What I tend to do is

– take the mean of the historical velocity (or throughput in stories/sprint)
– take the standard deviation of the same measure
– use that to make a probablistic forecast of what the team will deliver over time

That gives you a cone of uncertainty you can track if (say) you are talking about SAFe and capacity Vs delivery through a PI of 5+1 Iterations.

Actually, I’d do both throughput and velocity, as the team will learn something interesting 🙂

To make the forecast, you just add up the mean values, so

Sprint one, gives you 1 x the mean
Sprint two, gives you 2 x the mean etc

To find the expected range (ie the error bars above and below this) , you have to add up the standard errors. The standard error is the mean of the standard deviation. You can then get the range at sprint N by taking the square root of that sum.

So if SD is the standard deviation and SE is (SD) **2

Error bar at sprint 1 EB1 = SD
Error bar at sprint 2 EB2 = SQRT (2 x SE)
Error bar at sprint 3 EB3 = SQRT(3 x SE) and so on

So at the end of Sprint 3, there is

– a 16% chance the team will have done ( 3 x mean) + EB3
– a 50% chance the team will have done ( 3 x mean)
– an 85% chance the team will have done (3 x mean) – EB3

For planning, aim at that 85% bar (the low estimate) – you’ll have a buffer that allows you to pull more work, but just a 1-in-6 (ish) chance you’ll be wrong.

Works pretty well…