Estimate at completion (EAC) is a forecast of what the whole project will cost if we are honest about the remaining work. There is not one true formula. There are assumptions. Pick the assumption on purpose or the forecast will quietly assume the second half of the project will be a saint.
EAC = AC + BAC − EV — when the variance was a one-off
This formula says: keep the money already spent, then finish the remaining work at the original budgeted rate. Remaining work is BAC − EV. Add AC. In the example: $50,000 + $100,000 − $40,000 = $110,000. You overran by $10,000 on the work done; you claim you will not do that again.
Use it when the overrun had a cause that is finished — a one-off vendor bill, a storm, a mistaken purchase — and the remaining work looks like the original estimate. That is a strong claim. Write the cause down. If you cannot name why the past will not repeat, you are not using this formula. You are hoping with algebra.
Estimate to complete (ETC) under this assumption is BAC − EV = $60,000. EAC is just AC + ETC. If finance asks 'how much more will you need,' answer with ETC and the assumption, not with a shrug and a contingency shaped like a guess.
EAC = BAC / CPI — when this efficiency is the new normal
If the cost efficiency you have demonstrated is typical of the remaining work, the whole job will cost BAC / CPI. Here: $100,000 / 0.80 = $125,000. You will keep getting 80 cents on the dollar until the end. ETC is then EAC − AC = $75,000, not the original remaining $60,000.
This is the default when nobody can point to a structural change — no new team, no de-risked design, no removed scope. Past performance is the evidence. Teams dislike this formula because it is rude. That is a feature. A CPI of 0.80 that you 'expect to improve' without a named change is a novel, not a forecast.
If CPI is still unstable — week two of a ten-week project, or a single ugly invoice — do not treat this EAC as destiny. Say the sample is thin. A ratio from a small EV is a weather vane in a cupboard.
EAC = AC + (BAC − EV) / (CPI × SPI) — when time is costing money too
When cost inefficiency and schedule inefficiency will both persist — late work that also burns money, overtime, extended vendor retainers — divide the remaining budgeted work by the product of the indexes. CPI × SPI here is 0.80 × 0.80 = 0.64. Remaining $60,000 / 0.64 = $93,750. Add AC: EAC = $50,000 + $93,750 = $143,750.
Use this when being late makes the remaining work more expensive, not merely later. A delayed go-live that keeps a contractor team on the clock is the usual suspect. If schedule delay does not change cost (a fixed-price package already paid), this formula over-punishes. Do not use it to look 'rigorous.' Use it when the assumption is true.
Variance at completion (VAC) = BAC − EAC. Under the three assumptions you get −$10,000, −$25,000, and −$43,750. Put the assumption in the sentence: 'If the one-off overrun does not repeat, EAC is $110k. If current cost efficiency persists, $125k. If late-and-expensive persists, about $144k.' Then ask the sponsor which future they are funding. That is a decision. A single EAC with no label is a magic trick.
Summary & next steps
AC + BAC − EV assumes remaining work at the original rate. BAC / CPI assumes current cost efficiency continues. AC + (BAC − EV) / (CPI × SPI) assumes cost and schedule inefficiency both keep biting. Our example forecasts $110k, $125k, and $143,750. Name the assumption.
Next: reporting EVM — the same numbers in English a sponsor can use, without a formula sheet for a security blanket.