Performance analysis in practice: a full monthly cycle

0:00 One full monthly cycle, start to finish

So far we've covered each technique on its own. Now let's put them together, the way you'd actually use them in a real month. We're going to walk through one complete controls cycle for a fictional project: Desert Bloom Water, a desalination pump station upgrade in Saudi Arabia. Budget at completion, twelve million dollars. An eighteen-month plan. One million of contingency held separately. We're at the end of month nine. By the end of this lecture you'll have seen every step: validating the data, calculating performance by control account, finding the root cause, building a forecast range, comparing contingency with exposure, writing an executive headline with decisions, and logging actions. Use it as the template for your own monthly cycle.

0:52 Why practise the whole cycle?

Why walk through a whole cycle? Because techniques only create value in sequence. Perfect earned value calculations on unreconciled data are wrong. A great forecast with no root cause leads to the wrong action. A clear report with no action log changes nothing. Think of it like dominoes: each step has to knock over the next. And the cycle is only as strong as its weakest step. So as we go through Desert Bloom, notice how each step depends on the one before, and ask yourself which step is weakest on your own projects.

1:33 Step one and two: validate, then calculate

Step one, validate. The schedule is updated with actual dates and remaining durations at the end of month nine. Actual costs come from the ERP, plus four hundred and twenty thousand of accruals for work done but not invoiced. Progress is measured with weighted milestones for procurement and units with rules of credit for installation. Budgets in the schedule sum to twelve million, and ERP codes map to the WBS. Only now do we calculate. Step two: earned value by control account. Management, level of effort, CPI about nought point nine seven. Civil, nought point nine seven. Pumps and mechanical, nought point eight one. Electrical and controls, nought point nine three. Project total: CPI nought point nine, SPI nought point eight six. And here's the lesson: that total CPI of nought point nine hides the fact that one account is at nought point eight one and driving most of the overrun.

2:39 Step three: analyse causes

Step three, analyse. We interview the control account managers and check the evidence. For pumps and mechanical: the pump skids arrived with defects, rework on site used extra labour, and there's a vendor claim pending. But that's still a symptom. Why did defective skids reach site? Because the factory acceptance tests were waived to save time. That's the root cause, and it matters enormously, because it points to a preventive action for the remaining skids, not just a cost adjustment. For electrical and controls: it's behind mainly because it depends on the pumps, but its cost efficiency is acceptable. So the cause of electrical's schedule problem sits in another account. That's why analysis by control account, and by dependency, beats a project-level average every time.

3:33 Step four: the forecast range

Step four, forecast. For pumps: budget five million, earned value two point two, actual cost two point seven, so CPI about nought point eight one. Budget-rate EAC: two point seven plus the remaining two point eight, five point five million. CPI method: about six point one three million. Then the bottom-up: the rework is mostly complete, remaining installation is planned at near-budget productivity with a supervision upgrade, giving an ETC of three point one and an EAC of five point eight million. Controls selects five point eight, with a written rationale, and notes a pending vendor claim of up to two hundred thousand, excluded because it's uncertain. Project EAC comes to about twelve point nine million: within budget plus contingency of thirteen million, but with little margin. On schedule, the network shows a five-week slip; earned schedule suggests more; after checking a partial fast-track of electrical, the forecast is six weeks late at P50 and ten at P80.

4:42 Watch me do it: contingency versus exposure

Step five, risks and changes, and let me show you the check I never skip. In my sheet I have one cell for remaining contingency, one million. In the next, remaining risk exposure from the updated quantitative analysis. Here I'm using P80 minus P50 cost as the measure of exposure the project is carrying, one point three million. A third cell compares them. Exposure exceeds contingency, so it turns red and says escalate. That line goes straight into the executive summary. Then I check the change log. There's a client-requested SCADA upgrade, change twelve, plus three hundred and fifty thousand, not yet approved. And critically, the decision is time-sensitive: if it isn't decided by the fifteenth, there's a further three-week impact. That goes into the decisions section too.

5:38 Step six and seven: report, decide, act

Step six, report and decide. Here's the headline: forecast cost twelve point nine million, seven and a half per cent over budget but within contingency, and handover six weeks late at P50. Pump-skid defects are the main driver. Decisions needed: approve reinstated factory acceptance tests for the remaining skids, and decide on change twelve by the fifteenth to avoid a further three-week impact. Note that risk exposure exceeds remaining contingency. Step seven, act and follow up. Actions logged with owners: reinstate factory tests, the procurement lead. Submit the vendor claim, the commercial manager. The electrical fast-track plan, the planner. And an updated risk analysis next month, the risk analyst. Next month's review opens by closing out these actions.

6:29 Where monthly cycles usually break

Before we wrap up, let's be honest about where monthly cycles usually break, because I've seen every one of these. The first is calculating before validating: producing beautiful earned value charts on data that hasn't been reconciled, so a missing accrual quietly becomes 'improved performance'. The second is stopping at the symptom. 'Pumps are over budget' goes into the report, and nobody asks why the skids were defective, so the same thing happens on the next batch. The third is a single-number forecast with no range and no rationale, which invites an argument rather than a decision. And the fourth is the most common of all: no action log. The report is issued, the meeting happens, everyone nods, and next month the same variances appear with slightly bigger numbers. If you fix only one thing, make sure every review ends with owned, dated actions.

7:32 Lessons from the cycle, recap and try this now

What did this cycle teach us? Control-account analysis located the problem in minutes, where a project total would have hidden it. The root cause, waived tests, led to a preventive action, not just a cost adjustment. Multiple forecast methods framed a credible range, and the selected number came with a rationale. And comparing contingency with exposure triggered escalation early, while there was still room to act. So, to recap the cycle: validate, calculate by account, analyse root causes, forecast a range, compare contingency with exposure and review changes, report with decisions, then act and follow up. Your try-this-now: run the monthly checklist from the lesson on a real or practice project, and write a one-paragraph executive headline that ends with the decisions required.