Earned value management (EVM) can turn a project status report into an evidence-based cost forecast, but only when the measurement rules are credible. A chart that says a project is \u201c80 percent complete\u201d may describe expenditure, elapsed time or actual accepted work; these are not interchangeable. Forecasting begins by defining which work has earned its planned budget and which expenses have actually been incurred.<\/p>\n
This guide focuses on forward-looking decisions: selecting a forecast assumption, interpreting estimate at completion, testing the remaining efficiency needed and communicating uncertainty. It differs from basic earned-value reporting, which asks whether current spending and progress match the approved plan. The worked calculations are illustrative, not estimates for an actual contract or promises that statistical trends will continue.<\/p>\n
Before forecasting, establish the approved budget at completion (BAC), a time-phased planned-value baseline and a work breakdown structure that connects deliverables to costs. Planned value (PV) is the budgeted value of scheduled work at a reporting date, earned value (EV) is the budgeted value of genuinely completed work, and actual cost (AC) is the expenditure incurred for that work. All three must use compatible scope and cut-off dates.<\/p>\n
Suppose a hospital is upgrading its appointment scheduling system. The budget includes interface design, secure integration, training, acceptance testing and handover. A completed interface can earn its budgeted value when objective acceptance criteria are met. Paying a supplier’s deposit does not necessarily mean an equivalent amount of tested integration has been delivered. That distinction prevents spending from masquerading as progress.<\/p>\n
Progress rules may use completed milestones, weighted activities or measurable physical units. Whichever rule the team chooses, document it in advance and apply it consistently. A vague estimate from the loudest stakeholder creates volatility in EV, and an apparent improvement in a performance index may simply reflect changed measurement. An approved change to scope should be incorporated through disciplined baseline control, not hidden in retrospective progress scores.<\/p>\n
The basic historical indicators are cost variance CV = EV \u2212 AC and cost performance index CPI = EV \u00f7 AC. A negative CV indicates that accepted work has cost more than its budgeted value; a CPI below 1 indicates unfavorable cumulative cost efficiency. Schedule variance SV = EV \u2212 PV and schedule performance index SPI = EV \u00f7 PV compare earned progress with progress planned by that date.<\/p>\n
At a reporting point with PV of $250,000, EV of $200,000 and AC of $240,000, cost variance is \u2212$40,000 and CPI is approximately 0.833. Schedule variance is \u2212$50,000 and SPI is 0.80. These values describe different gaps: spending efficiency and the value of work accomplished relative to the baseline schedule. They do not by themselves identify which dependency or supplier decision caused the shortfall.<\/p>\n
Use the indices as questions, not diagnoses. Was work genuinely accepted later than planned, or was the original schedule unrealistic? Did rework add expense, or was a supplier invoice booked early? If the underlying data is unreliable, improving its collection may matter more than calculating another forecast. An earned value performance review<\/a> should connect those metrics to delivery evidence.<\/p>\n Assume the same project has an approved BAC of $500,000. If its cost efficiency remains close to 0.833 for all remaining work, a simple estimate at completion is EAC = BAC \u00f7 CPI, approximately $600,000. Estimate to complete is ETC = EAC \u2212 AC, approximately $360,000. Variance at completion is VAC = BAC \u2212 EAC, approximately \u2212$100,000. These are forecasts contingent on the stated assumption, not an authorization to exceed the budget.<\/p>\n The reason for dividing by CPI is that the historical delivery rate is being applied to the full budgeted scope. In effect the model assumes the team will continue earning about $0.833 of budgeted work value per dollar spent. The calculation is easy; the difficult question is whether the remaining tasks face similar efficiency, risks and resource constraints to the completed tasks.<\/p>\n A manager should explain the method alongside the value: \u201cOur current cumulative cost trend implies approximately $600,000 at completion if it persists.\u201d Reporting \u201cthe final cost will be $600,000\u201d hides the assumption. Report the current baseline BAC, an updated estimate, the remaining funding requirement and the date at which major project uncertainties may resolve.<\/p>\n Another standard assumption is that the one-time causes of the past overrun have been contained and the remaining work can be performed at its planned cost. In that case EAC = AC + (BAC \u2212 EV) = $240,000 + $300,000 = $540,000. The forecast is lower than the $600,000 CPI continuation case because it assumes cost efficiency on unfinished work returns to 1.00.<\/p>\n A third option when both cost and schedule inefficiency are believed to keep affecting remaining work is EAC = AC + (BAC \u2212 EV) \u00f7 (CPI \u00d7 SPI). With CPI about 0.833 and SPI of 0.80, this produces about $690,000. That assumption is severe: it compounds two historical inefficiencies. It should not be applied automatically or confused with BAC \u00f7 (CPI \u00d7 SPI), which is not algebraically equivalent to the full expression.<\/p>\n These scenarios create an understandable range rather than three equally true predictions. A supplier delay that is already resolved may favor the first recovery assumption; repeated quality failures in the unfinished integration may justify the trend or compound scenario. Explicitly state how material purchases, resource availability, the acceptance criteria and schedule recovery plan support the selected case.<\/p>\n An independent bottom-up estimate to complete is often valuable after major change. Ask each work-package owner to estimate remaining labor, procurement, testing and contingency from current facts, then add actual costs to obtain a new EAC. Comparing that figure with EVM forecasts can expose optimistic assumptions or atypical early work. The choice of project cost estimating method<\/a> depends on available evidence.<\/p>\n The to-complete performance index (TCPI) asks what efficiency unfinished work would require to meet a selected target. Against the original BAC, TCPI = (BAC \u2212 EV) \u00f7 (BAC \u2212 AC). Using the example, TCPI = $300,000 \u00f7 $260,000 \u2248 1.154. That means each remaining dollar would need to earn more than $1.15 of planned value to finish within the original $500,000 budget.<\/p>\n Compare this with the achieved CPI of about 0.833. A leap from 0.833 to 1.154 is substantial. A manager should not quietly assume it is achievable; the recovery plan needs a defensible explanation such as removal of substantial future rework, negotiated supplier savings or an approved scope change. Improving efficiency beyond the baseline is harder than \u201ctrying harder\u201d in a weekly meeting.<\/p>\n If the approved management target is instead an EAC of $600,000, a corresponding TCPI based on that EAC is (BAC \u2212 EV) \u00f7 (EAC \u2212 AC) = $300,000 \u00f7 $360,000 \u2248 0.833. The difference reveals what target the index uses. Label TCPI(BAC) and TCPI(EAC) clearly so a report does not disguise a revised forecast as performance against the original budget.<\/p>\n EVM schedule variance is measured in budgeted value, not days. An SPI of 0.80 does not mean the delivery date will be exactly 20 percent later than planned. The critical dependency path, resource constraints, remaining float and acceptance sequence determine actual dates. Near completion, traditional SPI can drift toward 1.00 even if the project finished late, because total EV eventually equals total planned value.<\/p>\n Use a critical-path or resource-constrained schedule and, where suitable, earned-schedule methods to estimate time. A missed interface dependency may block system testing while less important documentation work continues; the reported EV and the delivery date can move differently. Critical chain scheduling<\/a> explains how shared resource constraints and buffers change realistic completion expectations.<\/p>\n A forecast should also include nonfinancial quality evidence. Low test defect closure rates or unresolved acceptance criteria may imply extra cost even when the current CPI looks favorable. Conversely, paying for long-lead components early can temporarily depress apparent efficiency. Review the cause and timing before extrapolating a ratio to every remaining work package.<\/p>\n At each reporting period, reconcile the scope baseline, work acceptance, actual invoices and committed purchase orders. Ask the delivery team to explain major cost and schedule deviations, then update a small set of scenarios. Include the baseline BAC, current PV\/EV\/AC, CPI\/SPI, selected EAC method, estimate to complete and whether a change request is being considered.<\/p>\n Track committed but not yet accrued costs explicitly. A project may appear under budget if purchase orders or invoices lag behind physical completion. Conversely, an invoice can be prepaid before an associated milestone has earned value. Accounting and progress reporting should use a consistent recognition policy and present timing differences transparently rather than forcing them into a misleading indicator.<\/p>\n A useful management discussion might say: the integration team has completed only 70 percent of its accepted interface work, an external system now needs extra testing, and the forecast has moved from $540,000 to $600,000. The response is not simply to demand a higher SPI. It may be to negotiate test scope, allocate specialist capacity, protect acceptance criteria and request an authorized rebaseline if needed.<\/p>\n Forecasts should retain their version history: what was predicted last month, why assumptions changed and which actions reduced or increased the likely overrun. That history encourages learning and makes optimism visible. Separate the official approved baseline from the latest forecast; a new estimate does not itself authorize new spending.<\/p>\n An EVM forecast is most useful when it changes a decision early enough to matter. If the higher-cost scenario is plausible, the sponsor may need a funding contingency or a scope trade-off before the system enters critical testing. If better efficiency is supported by a credible plan, retain the lower scenario but test it against weekly earned-work evidence.<\/p>\n A health authority is migrating appointment systems across twelve clinics. Its baseline budget at completion (BAC) is $800,000, earned value (EV) is $300,000, actual cost (AC) is $375,000 and planned value (PV) is $400,000. The cost performance index is 0.80 and schedule performance index is 0.75. These indicators describe weak cost efficiency and lower earned progress than planned, but cannot explain whether data conversion, staff availability or a supplier delay caused the result.<\/p>\n Assuming this inefficiency continues, EAC = BAC \/ CPI predicts a $1 million final cost. If the overrun was exceptional and the remaining work can proceed at the original budgeted rate, EAC = AC + (BAC \u2212 EV) predicts $875,000. Those are scenarios with different assumptions, not alternative answers to memorize. The earned-value formulas<\/a> become useful when the underlying work measurements and remaining cost assumptions are credible.<\/p>\nCalculate a first estimate at completion<\/h2>\n
Choose the forecast that fits the future work<\/h2>\n
Test the efficiency required to meet a target<\/h2>\n
Understand what traditional SPI cannot predict<\/h2>\n
Build a forecasting conversation that changes decisions<\/h2>\n
Use ranges as management evidence, not certainty<\/h2>\n
Two defensible forecasts for a clinic data migration<\/h2>\n