EVM, or earned value management, is a method for viewing how well you are performing against your estimates.
Unfortunately it gets very little respect because everyone insists on making it more complicated than it needs to be. I have used a cut down version of EVM to demonstrate performance over time to great effect on several projects, which is very valuable for both reporting and control.
When should you use EVM?
EVM can be applied to any project provided it is tailored to the right level. The simple form presented here is good for an environment where you are doing a lot of projects of about one to two weeks in length.
This is an example chart from a five week project. It communicates very well that we kept closely to our estimate, with the exception of one week where our performance was higher than expected due to the discovery of a suitable library that reduced the amount of work required.
How do you calculate Earned Value?
A simple worked example is probably the best way to look at how it works.
We estimate that painting a bike shed will take 20 hours at £10/hour. It actually takes 40 hours at £11/hour. So, we are now 40 hours and £440 in and have one bike shed painted with an estimated value of £200 and 20 hours. We should have painted two bike sheds by now.
From this we get two interesting numbers:
- Our cost variance is £240 (£440-£200) per bike shed. It's costing 2.4 times as much as we thought.
- Our schedule variance is 20 hours (40-20) per bike shed. It's taking 2 times as long as we thought.
These numbers are interesting because they:
- Provide a feedback loop into our estimating process.
- Give an early warning sign that things are going bad. If we have 400 bike sheds to paint then we might be able to fix metrics like this before it turns into a real disaster.
- Track your efficiency in producing the desired output, expressed as a fraction of the input estimated in the plan.
You can see how the third point works by taking a closer look at those figures again. Cost and schedule variance are both calculated as a comparison of actual cost minus the progress we have achieved so far. In this case we have made one bike shed so we claim the estimated value of one bike shed, that is £200 and 20 hours, to deduct off the actual figures before working out the variances.
What should you do with the numbers?
There are two major outcomes to earned value, which are useful in different circumstances:
- Cost variance. Look at this if you are dealing with variable or controllable costs such as overtime or resources from a shared pool. If you have fixed costs within the scope of a project, such as a salaried development team, then you can ignore cost variance as outside your area of immediate interest.
- Schedule variance, or how you are performing against time estimates. This is applicable to any project with a deadline, but it can be calculated in several ways. This article uses the APM recommended input-hours-of-work but in the next article I will demonstrate how using elapsed time can make the data more useful for small teams.
We can plot the calculated multiples (in our example 2.4 and 2) at regular intervals. If we see a large movement in performance against the estimate we should be asking why this has happened and seeing which assumptions need to be changed.
A value of 1.0 works out to being exactly in line with the original estimate. The exact level of performance shown in this graph should now be more clearer.
How do you apply this to things other than bike sheds?
Extending EVM to something more complex than repeatable industrial processes is where things get slightly harder. However any project that can be estimated can have EVM applied to it. The less reliable the estimate, the greater the variation you should tolerate in your cost and schedule variances. There are two reasons for this:
- It is hard to estimate non-repetitive tasks. This is a problem that has yet to be satisfactorily answered, with reliable estimates still being rare. However the data you get from calculating EV on your project will help you refine your own estimating method.
- It is hard to say to what percentage tasks are complete. In practice the only method I have seen that reliably works is to subdivide tasks into manageable chunks that can be given unambiguous completion criteria, then claim the full value on completion. If you have too many tasks to comfortably work with on a small spreadsheet then possibly you should be segmenting the project into component phases, also with clear criteria for completion, and tracking one phase at a time. If your project is too large or complex for this approach then your project office should be able to help you with a computerized centralized planning system and a project controller to do the data entry!
Summary
EVM can provide a clear view of how well the work you are putting into a project is being spent. It does this by measuring your performance in terms of the the output that was required from your original plan.
It's key weakness is that it is only measures the things that are on your plan, and therefore some scheme to either update your plan or link earned value to business output is required. This is something I will address in the next article in this series.