I have been doing a lot of brushing up on EVM lately, and am ready to dig in again. Since I am already very familiar with the theory behind Critical Chain and buffer management, I am going to focus on EVM for awhile and then try to meld Simple EVM in the terms described by Joel Koppleman and Quentin Flemming to Critical Chain.
EVM at it’s core is a way to measure performance in terms of cost, schedule, and technical performance. One of the key benefits I have heard cited regularly is that with EVM, you can tell early on whether or not a project is at risk. Quentin Flemming said in an interview on The PM Podcast that there is strong evidence that poor EVM performance at the 20% done mark in a project is very unlikely to be made up unless more resources are added, more funds are appropriated, or the scope gets trimmed.
Another benefit of EVM is objectivity. It measures performance compared with an original baseline schedule and budget. The performance indexes in EVM allow comparisons between projects even when their actual budgets and schedules are very different. You can compare a 6 month project to a 6 year project in the same terms. Because of the objective and universal measure, it also can help make project performance visible to all levels of stakeholders.
Why Not EVM?
There are a few criticisms of EVM I have heard that I would like to discuss. First, full-blown EVM is simply too much overhead for small projects. Simple EVM was introduced for this very reason, and I believe it could be a suitable replacement for smaller projects.
Another thing to keep in mind is that EVM does not take the critical path into consideration. It treats all work equally, which works really well for cost performance but not so well for schedule performance. At first blush, it seems to me that in addition to the SPI (schedule performance index) which I will detail in later posts, there should be an index that tracks progress along the critical path. This could be done by simply tracking schedule performance on the critical path against the original baseline. In this way, the SPI relates more to the amount of work done than the timeliness of the project. They are different but related ways of looking at project schedule performance.
PV - Planned Value AKA BCWS - Budgeted Cost of Work Scheduled
- This is a time-specific estimate only. What was planned to be spent at this time if the schedule was perfect?
EV - Earned Value AKA BCWP - Budgeted Cost of Work Performed
- This is a task-specific estimate only. What was planned to be spent on tasks actually completed right now?
AC - Actual Cost AKA ACWP - Actual Cost of Work Performed
- This is reality. What did you actually spend? Make sure the method for auditing cost matches up with how you made the original plans (did you include indirect costs, or just direct, etc?)
Variances are always expressed in currency. They are just what they sound like, the difference between where you are at and where you had planned to be.
CV - Cost Variance
- CV = EV - AC (or CV = BCWP - ACWP)
- Tasks 1-20 are done. You planned on that costing $90,000, but it actually cost you $85,000. $90k - $85k = $5k
SV - Schedule Variance
- SV = EV - PV (or SV = BCWP - BCWS)
- Tasks 1-20 are done, but you planned on getting 1-23 done. 1-20 was planned to be $90,000, and 1-23 was planned to be $100,000. $90k - $100k = -$10k
Indexes are used to measure performance on a ratio basis which could actually be compared across projects regardless of their comparative sizes. If you are working perfectly to plan, your indexes will both be 1.00. Greater than 1.00 means you are doing better than planned, less than 1.00 means you are doing worse than planned.
CPI - Cost Performance Index
- CPI = EV / AC (or CPI = BCWP-ACWP)
- Using the numbers above, $90k / $85 = 1.06
SPI - Schedule Performance Index
- SPI = EV / PV (or SPI = BCWP / BCWS)
- Using the numbers above, $90k / $100k = 0.90
So, looking at these variances and indexes, you can see in the example above that we are behind schedule, although it looks like we’ve managed to keep our costs down so we are actually below budget too.
EVM is meant to be used during project execution to monitor progress and hopefully find problems early so that adjustments can be made. Forecasting is an important part of finding out how big of a deal your CPI or SPI is at this point in the project.
Using the same numbers from above, so:
- PV = $100k
- EV = $90k
- AC = $85k
- CPI = 1.06
- SPI = 0.90
BAC - Budget at Completion
- This is the total planned budget for the project (no EVM calculations)
- Let’s say we estimated the total cost for this project to be $200k.
ETC - Estimate To Complete
- How much is it going to cost to finish this project?
- 3 ways to get this (choose the best one for your situation):
- New estimate - go and re-estimate the work remaining. Most accurate, most time-consuming.
- EVM based on atypical variances - past performance is expected to change, use ETC = BAC - EV (or ETC = BAC - BCWP) $200k - $90k = $110k
- EVM based on typical variance - past performance is expected to continue, use ETC = (BAC - EV) / CPI (or ETC = (BAC - BCWP) / CPI) ($200k - $90k) / 1.06 = $103.8k
EAC - Estimate At Completion
- This is simply what the total project is expected to cost given what you know right now
- Use the ETC from whatever method you chose to use.
- EAC = AC + ETC (or EAC = ACWP + ETC)
- So for atypical, we would say $85k + $110k = $195k
- For typical, we would say $85k + $103.8k = $188.8k
The next article in this series will discuss some examples.
Josh Nankivel is the Vice Chair of Special Projects for the Students of Project Management SIG of PMI, and a project management student/enthusiast. His website is http://www.pmstudent.com.
No comments yet.