Earned value is a topic that makes many roll their eyes in an expression of exasperation. Yesterday’s silver bullet. Like all powerful ideas that are picked-up by the ‘instant-fix-brigade’ they are more misunderstood and over claimed for than is appropriate. Use correctly it is unsurpassed as an indicator of status.
My first claim against EV is that its expression of value is totally confusing to the vast majority. My second is that its formulas give an illusion of reliability and third its results are utterly useless without a solid notion of what constitutes “done”.
In an agile world most EV is a Story with out acceptance criteria, in a traditional world its a requirement without link to product or process specification in the QMS (quality management system). As soon as the question is asked “how do we tell if we are ‘done'” the real difficulties that underlie every endeavor to create results for someone else are exposed.
A customer safe answer is “we are done when we agree we are done”. In earned value this is “no credit due until 100% complete (against the requirements as I see them today and no matter what I said before)” – Clearly untenable for most suppliers seeking to submit a final or worse yet interim claim for payment without deduction or dispute.
EV’s real value is in the thinking it forces in how to recognize both “done” and “part done”. Wise insight ties part-done to process standards and to the core of what should be in estimating models (more on estimating another day – our complete Agile Software Estimating course is in the downloads area).
To implement EV a set of ‘EV-Types’ must be applied to all outcomes/ results/ impacts/ products/ deliverables during the design of te baseline and project controls: the sad fact is most people guess a percent-complete during execution and then anything calculated by EV’s formulas is utterly worthless – a complex and expensive way of living with 90% complete syndrome.
At its heart EV makes two comparisons, expressed either as a variance (using subtraction) or a ‘performance index’ (by dividing to express a ratio)
- What has been achieved* compared with what what scheduled to be achieved by now (Yielding a schedule variance and schedule performance index SV & SPI)
- What has been achieved versus what resources have been consumed
Yielding Cost status – CV & CPI)
*the expression of achievement (classically in US-Dollars) can be in any quantity that shows what the base-lined budget was, so If I am to build a 200 brick wall, at a linear rate of progress across 360 minutes at a cost of £120gbp then 54 bricks, 97 minutes and £32 are all equivalent expressions of achievement. They all equate to 27% – the 27% is derived FROM the 54 bricks or “97 minutes worth” not the other way around!.
In a task with a physical deliverable made of uniform and atomic items like a brick-wall this is easy – 54 bricks laid! if the task is intellectual – EG design an electrical system whose correctness is proved only once the installed switch is flicked then the best assessment for the customer is 0% until that switch instantaneously shows 100% is reality.
In all cases EV should accumulate and report costs as they are incurred. The time-phased elements of the ‘plan’ are only relevant to assessments of schedule. The actual, demonstrable achievement by the team is relevant to both schedule and cost calculations.
A ‘reasonable’ set of EV types is
0 – 100 in which costs accrue as incurred but no progress is acknowledged until a working result can be demonstrated. With a Product Breakdown Structure (or Work Breakdown Structure that properly honors PMI’s definition [and I’ve yet to see one that does even in PMI publications]) reasonable granularity may be achievable but not until after “Detailed design” or each sprint planning meeting (after which EV’s SPI is called “velocity” – same concept different jargon). 0-100 is perfect for tasks whose baseline performance is to start and end in the same reporting period.
50 – 50 is a compromise over 0-100’s often unfair mis-representation of the supplier’s efforts. In a 50-50 arrangement 50% of the total achievement is credited the moment work starts but nothing more is credited until completion. It is two wrongs trying to be a right and it sorta works if used for tasks spread evenly across just 2 reporting periods.
20/ 80 or 33/ 67 or any other agreed split is then a compromise on the compromise that seeks to accommodate uneven scheduling across two reporting periods.
Percent complete is classic (and normally misleading) but reliable if the PBS and WBS are granular enough AND estimating returned a transparent Basis_of_estimate AND the work delivers entirely physical results E.G. that electrical design calls for installation of 366m of cable all of which is equally easy to install…
Weighted Milestones is an attempt to balance the strengths and short-comings of the above and Capped-Weighted-Milestones (aka Percent-Complete with MS Caps) is an even better balance of strengths and weaknesses of the other methods. In Weighted Milestones a review process is defined and the successful exit of a review (by independent expert peers for example) allows claim of the %age agreed at planning time. Until the milestone is reached only the value of the previous milestone (or zero) is claimable – a sort of variable 0-100 with intermediate steps.
Capped-Weighted-Milestones allows the supplier to claim their assessment of percent-complete (hopefully calculated by an audit-able process tied to the basis-of-estimate) but no more than the value of the next as-yet un-passed review. This is perhaps the method of choice for most projects under contract where the effort to monitor the contract is not a significant consideration; the supplier can show the granularity and continuous progress in percent complete terms while the customer cannot suffers run-away enthusiasm and optimism from the supplier in excess of the next review point.
There are two more EV Types that must be mentioned. One where two tasks are bound together such as project management and technical work. If the team are running fast (or slow) on the technical work then half way through the schedule the PM is not 50% complete. This method is called Apportioned Effort and the dependent tasks “claim” the same %complete CALCULATED for the independent task: this mechanism is (always? and only?) used for monitoring tasks. The last EV-Type is always and ONLY appropriate to arrangements tied to the passage of time like hiring a car. As the clock ticks so value is ‘earned’. This EV-Type never applies to provisions like project management that run to delivery of a result!
EV’s value is the consideration of EV-Types that are best applied to every result in the product-backlog or deliverable on the PBS and task in the WBS’ work-packages and that is all impossible without a description of “done right and proper” for every task and for every result whether physical artifact or service/ intangible result. So; how would *you* gauge your progress partway through a corporate culture change!? 🙂