Indexes turn the same three numbers into efficiency. A variance says how large the gap is. An index says how well you convert spend and time into earned work — which is what you need if anyone is about to ask whether the rest of the project will behave.
CPI: value earned per dollar spent
Cost performance index (CPI) = EV / AC. In the example: $40,000 / $50,000 = 0.80. You are getting 80 cents of budgeted value for every dollar you spend. A CPI of 1.00 is on plan. Above 1 is better cost efficiency; below 1 is worse. This is not a vibe. It is a ratio.
Sponsors understand 'we get eighty cents on the dollar' faster than they understand −$10,000 CV, because the ratio scales. The same CPI on the next $50,000 of spend is a forecast, which is why the next lesson exists. Do not call 0.80 'broadly on track' unless you enjoy being asked to define 'broadly' in a steering pack.
CPI can lie if EV is inflated — percent-complete guesses, partial credit for work that is not done, or a baseline that already hid contingency in every package. Garbage EV divided by honest AC is still garbage. Protect the definition of done or the index becomes interior decoration.
SPI: value earned versus value planned
Schedule performance index (SPI) = EV / PV. Here: $40,000 / $50,000 = 0.80. You are earning value at 80% of the planned rate. Same rule: 1.00 on plan, above 1 ahead of the plan's earning rate, below 1 behind.
SPI shares SV's blind spot. You can have a respectable SPI while the critical path is dead. You can have a weak SPI because you scheduled a mountain of work in a month the team was never going to swallow — a planning error wearing a performance costume. Read SPI, then read the network.
Near the end of a project SPI tends to crawl toward 1.00 because almost everything planned has either been done or re-baselined. That late-project hug is not a recovery miracle. If you need schedule insight in the last stretch, look at remaining duration on the critical path, not at an index that is running out of unfinished planned value to compare against.
Use the indexes; do not worship them
A useful pair: CPI for money efficiency, SPI for earning rate, both dated. 'CPI 0.80, SPI 0.80 at week five of ten, BAC $100,000' is a status sentence. 'Indexes are a bit soft' is a weather report. Add the trend if you have two or three status dates — a CPI falling from 0.95 to 0.80 is a different story from a stable 0.80 you already funded.
Do not average your way to comfort. A workstream at 1.20 and another at 0.50 is not 'about 0.85, so nearly fine.' The second workstream is a project. Report the aggregate and the offender.
Indexes below 1 are not a moral failing and indexes above 1 are not a personality. They are prompts: why, is it typical, and what will we change? If nothing will change, you are collecting decimals. The next lesson is what you do with them when someone asks what the project will cost.
Summary & next steps
CPI = EV / AC. SPI = EV / PV. Our example is 0.80 and 0.80: eighty cents on the dollar, and 80% of the planned earning rate. One is not 'on track.' Protect EV quality or the ratios are theatre.
Next: forecasting EAC — three formulae, three assumptions, and the damage done when you pick the cheerful one by habit.