Earned value for planners: SPI, SPI(t) and when to trust them

Earned value tends to live with the cost team. But if your programme is resource- or cost-loaded, you already have everything it needs: a budget on each activity, a percent complete, and planned dates. The planner is well placed to read it.

The three numbers#

• Planned value (PV): the budget of the work that should have been done by the data date, on the planned dates.
• Earned value (EV): the budget of the work that has been done, which is each activity's budget × its percent complete.
• Actual cost (AC): what the work done actually cost.

From these come the two indices everyone quotes: SPI = EV ÷ PV, and CPI = EV ÷ AC. Below 1.0, you're behind plan or over budget.

The trouble with SPI#

Cost-based SPI has a well-known flaw. As a job nears completion, all the planned work is eventually done, so EV climbs to meet PV and SPI drifts back towards 1.0, however late the job is. A project that finishes six months late reports an SPI of 1.00 on its last day.

That makes SPI least useful exactly when people are watching it most closely.

Earned Schedule#

Earned Schedule asks a different question. Not "how much value have we earned?" but "when should we have earned it?".

Take the value earned so far and find the date on the planned value curve when that value was due. That's the earned schedule, ES. Compare it with the actual time elapsed, AT:

• SV(t) = ES − AT, a variance in working days, not pounds.
• SPI(t) = ES ÷ AT.
• IEAC(t) = planned duration ÷ SPI(t), which gives a forecast finish date.

SPI(t) doesn't drift back to 1.0. A job running at 88% of the planned rate stays at 0.88 until the day it finishes.

Two forecasts, one question#

On a live programme you'll usually have two forecast finishes. The critical path assumes the remaining work runs at its planned durations. Earned Schedule assumes the job keeps performing as it has so far.

When they disagree, and they often do, the gap is the conversation to have: is there a reason to expect the rest of the job to run better than the first half?

What to watch#

• Earned value only means something if the budget is on the work. A programme loaded only on a few activities gives a misleading EV.
• Percent complete must be real. Physical percent complete that nobody updates will flatter EV.
• Measure against the right plan. If the export doesn't carry a baseline, the planned dates are the baseline. Know which one you're using.