James,

A fascinating example of (one of the reasons) why the way that EV is applied to schedule is problematic.

What you are describing happens all the time with EV (i.e., getting BCWP credit for BCWS costs that are not yet in the denominator). The only reason that its particularly egregious in this case (to the point of causing a mathematical error!) is because it’s happening right at the start of the project, when there’s nothing else in the denominator, either. When it happens a couple of months later (as it frequently does!), it might just result in an SPI of 1.20, which most people would happily accept, despite the fact that it may have zero real implications for schedule performance.

First: the standard, accepted application of EVMS will give you an SPI of infinity. Whether your customer likes it or not, that’s the math. If submitting that as your report, I would show the math, and buttress it with a raw SV report (+ $100,000) and a statement that "if the current BCWP dollars were adjusted earlier in the BCWS to the actual start of work, then the current SPI would be 1.00 (or whatever)."

Second: Is the work that has been advanced on the critical path? Because if it’s not, it makes absolutely no difference to the project duration. The CP determines the project duration, so even though doing work off the CP ahead of schedule jacks up the SPI, it’s merely a metrics distortion that does not affect the project duration (unless it thereby frees resources to work earlier on CP activities, but that’s another kettle of fish). This is the reason that I always recommend that, for schedule tracking, the BCWS be on the backward pass (LS and LF dates) where there is no float distortion, and credit not be allowed for work done ahead of schedule. But my recommendation is not the standard way.

BTW, in the event that the work you’ve done IS on what was (and still may be) the CP, I’d try to show this by a DRAG Cost report: how much the "true cost" of each of the CP activities, and ultimately of the whole path, has been reduced by starting the work earlier, and thus pulling in the expected duration.

BTW, is this an outage project (I’m thinking not!)? If so, pulling in the CP even as much as two days must be worth millions, far more than any EV report is going to recognize. If this advance is on the CP (or, again, frees up resources to advance the CP), I urge you to find a concrete way to quantify the credit you truly deserve!

Frank Neal wrote:

"Hi Stephen, I have not read the book, but would love to know what the terms DRAG and CLUB relate to. I apologise if you have explained elsewhere in the forum and more than willing to go to the specific thread if there is one."

Sorry to be slow getting back to you on this, Frank. In addition to some health issues, I have also been busy helping a client whose customer wants to pull in the completion date from January of next year to November of this year. They’ve sent me the current schedule (only about 600 activities) and asked me to identify how much time each CP activity is currently adding (i.e., its DRAG), so they can get the "most bang for the buck" pulling it in. This is exactly how DRAG should be used and, frankly, the analysis is a helluva lot more fun that watching my Windies team get stomped by South Africa in the Cricket World Cup!

You’ll find, I think, a fairly comprehensive discussion of DRAG in the separate Critical Path thread in this subforum. I explained it posts 37, 44, 46 and 47, I believe. I also referenced some articles:

http://www.chiefprojectofficer.com/article/135

http://www.projectsatwork.com/content/articles/234282.cfm

And other articles, webinars, etc. referenced at my website:

http://www.totalprojectcontrol.com/about/articles/articles.html

Additionally, there is an exercise in DRAG computation in a simple (FS only) network under the Exercises tab at the site:

http://www.totalprojectcontrol.com/exercises/exercises.html

Warning: The Projects@Work articles require registration, but it’s free. The two older Webinars are archived by the respective PMI SIGs, but I believe they cost $10 or something if you are not a memeber of the SIG. The newest Webinar "The Cost of Tripping on the Critical Path!", is being broadcast live tomorrow (Thursday) at 1 pm Boston time (EDT USA). It costs $5, but if anyone here wants to tune in, send me a personal message in the next six hours or so and I’ll send you a promotional code to get it for free. You can access the info and register with the code at:

http://www.met-sig.org/products.aspx?category=68&product=301

I’ll deal with the CLUB in a separate post.

Anoon,

I know it’s hard to believe, and I hate to be the one to burst your bubble, but I have evidence:

Buildings collapse.
http://archive.salon.com/books/review/2000/07/13/wearne/

Airplanes fall.
http://www.esdjournal.com/articles/Lightning/russian%20crash.htm

Boats sink.
http://www.kltv.com/Global/story.asp?S=6331334

And I spent an extra 45 minutes last weekend driving around downtown Boston looking for a hotel because my GPS system kept sending me down exit ramps that had been permanently detoured a few months earlier.

Actually, this latter tends to happen a lot more often than the greater catastrophes. I wonder why that is?

Could it have to do with risk mitigation? After all, a lost hour of my time is probably worth a tiny bit less than even the two lives lost on the Sea Diamond, never mind the 114 in the Kansas City Hyatt. How do you think they mitigate risk? By going to an extended level of caution and "quality control", do you think?

Yes, I’d guess it’s possible to have estimates of effort on activities that are accurate more than 99% of the time: just take every estimate and add a bunch of standard deviations. Or, simpler, just add a single zero at the end. If a 50-50 estimate seems like 2,000 manhours, change it to 4,000, or 20,000, and you’ll probably exceed it less than 1% of the time (only probably because of Parkinson’s Law).

Actually, Anoon, I can now see the error of my ways. You have shown me the Antarctic light! I would hereby like to recommend to all readers of PP that, whenever they are bidding on a contract, they routinely bid 10 times more than they think they will need.

And the beauty is, one contract like that and you can sometimes retire!

Anoon wrote:

"Value is a generic term, depending on the unit of measure that has been agreed (may it be in cost or man-hours etc.)"

I guess I’m not sure what you mean by "a generic term". And far be it from me to argue with the Red Queen’s proclamation that: "a word means what I want it to mean - neither more nor less." However, I use it to refer to the noun "value" per the first three definitions (and particularly #3) in the Fourth Edition of the American Heritage Dictionary of the English Language:

"1. An amount, as of goods, services, or money, considered to be a fair and suitable equivalent for something else; a fair price or return. 2. Monetary or material worth: the fluctuating value of gold and silver. 3. Worth in usefulness or importance to the possessor; utility or merit: the value of an education."

This is the "value" the expectation of creating which makes a customer/sponsor decide to invest $X of "cost". The DIFFERENCE between this expected "value" (however gnerated) and the cost is the raison d’etre of every project, and therefore is definitely NOT the cost.

Anoon also wrote:

"IMHO, the accuracy of EV lies on the level of details."

The understanding and use of EV lies in recognizing that the purpose is not accuracy, but rather a measure of the degree of budgetary INaccuracy. EV was developed by the U.S. DoD because they’d give a contractor $150 million to do a project, and after a year $149 million would be spent and 10% of the work would have been done. EV broke the budget into weighted milestones so that if the first $20 million was spent on milestones that were budgeted for just $15 million, the DoD could get a sense that the project was "trending" toward being $50 million over budget. (More detailed analysis of where the overspending was occurring might ameliorate or exacerbate those numbers.)

Accuracy is a very overrated (and highly unachievable!) quality of project planning. Project control techniques like EV are designed to start measuring the inaccuracies of the plan, analyzing the trends, and creating updated expectations based on actual performance. In the words of General Dwight Eisenhower: "In preparing for battle I have always found that plans are useless, but planning is indispensable." EV analysis makes future predictions based on the variance between performance and plan, planning still having been an indispensable process even when the plan is highly inaccurate.

Anoon also said (in the Critical Path thread):

"one thing for sure! if there is no Schedule, there is no Critical Path!"

This is patently false, and a very unfortunate misapprehension under which project managers/planners who do not understand critical path method frequently labour. Every project, whether using critical path analysis, resource-loaded schedule, or making no attempt to schedule at all, has a critical path.

If a project has three activities (A = 40, B = 30, C = 30) arranged logically as FS relationships, and a fourth activity (D = 10) with no logical predecessor nor successor, the longest path (and the project) SHOULD take 100. However, if the true nature of the other path is:

Delay in starting D = 95
D = 10

then the true (ABCP!) critical path, and the project duration, will be 105. This will remain true whether or not any attempt was made to develop any project schedule whatever -- the only difference being that developing a CPM schedule would have shown the planner that "Delay in starting D" could not be more than 90 without it becoming critical and lengthening the project duration.

Further, the critical path, as Charlie has said, is of huge importance in that:

1. It ALWAYS determines the length of the project,
2. Which in turn impacts how much the value will be, when it will be created, how much the cost will be (through Marching Army Costs), and, most important,
3. How much "profit" will be generated through the project investment.

(Bearing in mind that, in a contract situation, the actual terms may distort these numbers into being very different for the customer vs. the contractor!)

Ultimately, the project manager/planner is certainly free to ignore any and all scheduling. But that does not make the critical path go away -- it only implies that it will be considerably longer than it might need to be! The critical path is there in every project -- you can plan, optimize and manage it, or not. But any project manager/planner who does not understand the intimate details and value (yes!) of critical path analysis is a charlatan.

Anoon also said: "Stephen, Further, this is the relevance of defining and estimating an Activity Duration accurately. It has nothing to do with DRAG or CLUB (Night)."

I’m sorry, I thought the thread was about earned value, not activity durations. One is cost, the other is schedule, and for whatever reason, project control tries to conflate the two.

Making schedule projections based on a cost metric is like asking: "If the 18-hole golf course has a par of 72, and a player has shot 5,6,4,5 for the first four holes which had pars of 4,5,3,4, how many hours and minutes will it take him to play all 18 holes?"

Actually, it’s even worse than that, because as I said in another thread, schedule is ruled by the critical path (yes, whether or not the project was ever scheduled!), and earned value shedule metrics (SV and SPI) do not differentiate between critical path activities and activities with float. (I also discussed how to correct for this.)

Certainly, I am all for estimating activity durations accurately. (And actually, I believe I read somewhere that it did happen once, on a small project back in ’71, I think -- don’t remember the details, though). Usually, many durations will vary from estimates to greater or lesser degree. For scheduling purposes, what’s important is to be able to see, and see quickly(!) what the impact is on the critical path, on the possible "migration" to a new CP, on the changes in DRAGs and DRAG Costs of future CP activities, and on the "windows of availability" for future resources.

This last is, of course, only tangentially related to their CLUBs. In general, I have tried not to take issue with your incorrect statements in these threads (the one about there being no critical path without a schedule was posted over a week ago). I certainly don’t want to argue with you, and you are free to ignore my input.

But I do know a bit about both scheduling and earned value. And if you want to denigrate critical path analysis and ridicule terms such as DRAG and CLUB, at least have the goodness to first read the book where they are explained.

Since we’re getting into a rather detailed discussion of earned value, I think it’s important for me to repeat something that so many people (including PMI!) tend to forget: earned value is NOT about value -- it’s about cost!

"Using dollars, you may have a task like install expensive equipment, so little EV until last minute then large EV when installed."

Yes, this is sometimes called "The Purchased Items Distortion" -- the fact that materials-and-equipment intensive activities, due to their large budgets, can sometimes have an impact far greater than their likely volatility or risk. Labor intensive activities usually tend to be more "volatile".

"other is manhours, which is the reverse of above, as very little man hours spent on installing expensive equipment.
say a few men and a crane for lowering into position."

Weighting by manhours is considered the "remedy" for the Puchased Items Distortion.

"We are also measuring Engineering, procurement, as well as construction but the problem is when we are reporting say 30% complete, they go to site and see nothing there!."

You didn’t say whether you were using milestones or continuous activities. If you’re reporting EV within continuing activities ("This activity is 60% done, so I get credit for 60% of it’s EV."), I’m sympathetic to their position. That’s why I suggested 1/3 - 2/3 milestone weighting system. If they don’t like that, try negotiating 25% - 75%, or whatever.


Are they more concerned with schedule or cost? If schedule, then I’d also suggest two further stipulations:

1. That no milestone will be considered achieved until the period in which it is scheduled to occur.
2. That the baseline for schedule data (BCWS) be the backward pass dates (ALAP) of the critical path schedule. That way, there is no distorting float on the milestones in the BCWS (everything is critical!)

BTW, I think those two suggestions should always be used for EV tracking of schedule.

Trevor, sorry to be slow responding. But you made several thoughtful points, and I wanted to be able to take the time to respond equally thoughtfully.

Trevor wrote:

1. "Stephen, wouldn’t you say that BCWS, BCWP and ACWP are all mis-named and mis-interpreted in two ways?... Second, they should be BCTS, BCTP, ACTP, ie T=Task, since it is about measuring the progress of Tasks, not Work (Hours) (as these terms are applied esp in MSP)."

Trevor, I agree that I’d prefer those terms, in that they more accurately reflect the essence of the purpose and application. But to me, that’s a much less important battle than the dangerous conflation of "cost" with "value", a problem hugely exacerbated by the 2000 PMBOK Guide’s changing the name of the BCWS to "Planned Value (PV)"!

2. "First, it is of course about cost, not value (whatever that is)."

Whatever it is, and whatever is generating it, it’s the whole reason someone’s spending (i.e., "investing") that cost and expending the effort and time, on EVERY PROJECT! That seems like a pretty important reason fror not being confused about it.

And however much it might actually be, the one thing we can be sure of is that SOMEONE (customer, sponsor, employer) is expecting the value to be greater than the cost! And the size of the difference between those two metrics (a difference primarily dependent on the interaction between scope, schedule and cost) is "project profit" and should be THE most important metric of the relative (as opposed to binary!) degree of project success.

3. "There are lots of different suggestions about which Costs to include or exclude from EV calculations, for all sorts of different reasons, but shouldn’t we start from a proposition that every last cent (Work Costs, Materials and Equipment Costs, Fixed Costs) that can be associated with a Task should be included?"

Not in my opinion. For determining project budget/actual cost, absolutely! But from the purpose of project control, the purpose of earned value controls is to be a predictor of future performance. As such, the method that will produce the EV metric with the best predictive value should be used.

Some costs (i.e., labour), and the costs of some activities (labour-intensive, design, re-work), are often more "statistically significant" than others (i.e., equipment). This is one reason why, in making future cost projections, it’s often important to compute EV by functional area -- if Design worked at a CPI of 0.50, but all the Design work is finished, to "straight-line" that 0.50 CPI into the project cost ETC without recognizing its distortive potential is being deliberately obtuse.

If I may use a cricket analogy, the way that cricket batting averages are computed is one of my pet peeves. They do not include not outs in the denominator, essentially making the assumption that the runs scored in such innings are "magically" accumulated, and added to the numerator! Thus Steve Waugh’s and Brian Lara’s batting careers look similar: Waugh at 51.06 and Lara at 52.88.
However, Waugh had 46 not outs in 260 Test innings! Compared to Lara’s 6 in 232!

One can say whatever one likes about these two as players, captains, etc. But there is no question that when the two walked out to the middle, there was a clear expectation that Lara would score more runs. For the averages to have any predictive value, they should be computed like earned value milestones with 50-50 weighting -- .5 for each innings started and .5 for each out. This would lead to a Waugh average of 46.11 vs. 52.19 for Lara.

4. "Isn’t part of the ’distortion problem’ just a matter of when to accrue the Costs?"

No, although that’s certainly an important potential distortion, too (the "Disbursement Delay" distortion, to be precise). But for purposes of predictive value (which EV is really about), the distortion based on the volatility of different cost types is very real.

(Of course, schedule metrics like SPI potentially incorporate the much greater "distorter" of float, by ignoring the overwhelming importance of the critical path. I discussed how to avert this in my earlier post.)