20.2 Variance Reports & Explaining Project Financial Health
Key Takeaways
- A variance report must show budget after approved changes, earned value, actual cost including accruals, both dollar and percentage variance, root cause, forecast impact at completion, and an owner-assigned action.
- Variance thresholds should require both an absolute dollar test and a percentage-of-budget test, so trivial dollar amounts and small percentages on very large accounts do not consume review time.
- A cause column that says 'poor productivity' restates the variance; a real cause names an event, a quantity, and a responsible interface, which is what makes the action column writable.
- CPI and SPI read as a quadrant: unexpectedly favourable early CPI is far more often missing accruals than genuine efficiency, and favourable SPI with adverse CPI usually means acceleration was bought.
- A forecast must always be reported with its method named, because the typical, atypical, and composite EAC formulas can differ by tens of millions on identical data.
20.2 Variance Reports & Explaining Project Financial Health
Blueprint tasks 3.H prepare variance reports, 3.D interpret CPI/SPI, and 3.B prepare reports to explain financial health of the project to many types of stakeholders cover the analytical heart of Domain 3. A variance report that lists numbers is data entry; a variance report that explains causation and prescribes action is cost engineering. This section also carries Domain 4's 4.I validate cost analyses — before a variance explanation reaches a stakeholder, the analysis behind it has to be checked against the source cost data, the progress basis, and the accrual cut-off.
1. Anatomy of a Variance Report
| Column | Content | Common defect |
|---|---|---|
| Control account | Identifier and description | — |
| Budget (BAC) | Current approved, after approved changes | Original budget shown, so approved changes look like overruns |
| Earned value (EV) | Value of work performed | Computed by an inconsistent rule |
| Actual cost (AC) | Cost incurred including accruals | Invoice-only, understating AC (see 17.2) |
| CV and CV% | EV − AC; CV / EV | Percentage omitted, so a large account's small percentage looks alarming |
| SV and SV% | EV − PV; SV / PV | Read as a schedule position (see 19.2) |
| Cause | The specific, evidenced reason | "Poor productivity" — a restatement, not a cause |
| Impact at completion | Forecast effect, in dollars | Omitted, so period variance is mistaken for final impact |
| Action and owner | What is being done, by whom, by when | Missing, so the report informs without changing anything |
Thresholds
Reporting every variance equally buries the ones that matter. A workable rule sets two tests, both of which must be met: an absolute value threshold (say $50,000) and a percentage threshold (say 5% of the account's budget). This prevents both a trivial dollar variance on a small account and a small percentage on a very large account from consuming review time — while ensuring anything that is both materially large and proportionally significant is always examined.
2. Root Cause, Not Restatement
The single most common weakness in variance reporting is a "cause" column that restates the variance. Use the decomposition from Section 18.3 and then push one level further.
| Restatement (not a cause) | Actual cause | Action it implies |
|---|---|---|
| "Labour overspend" | Erection productivity 1.14 vs 1.00 planned because engineering issued steel drawings 6 weeks late, forcing out-of-sequence erection | Recover engineering; re-sequence; quantify disruption for change |
| "Material cost higher" | Alloy piping bought at spot price after the framework agreement lapsed | Reinstate framework; price remaining requirement |
| "Subcontract overrun" | Scope gap between the civil and mechanical packages, executed as day works | Close the gap by variation; recover from the responsible package |
| "Schedule slip" | Permit issued 5 weeks late, delaying access to the work face | Claim time; assess acceleration cost against liquidated damages |
Each real cause names an event, a quantity, and a responsible interface. That is what makes the action column writable.
3. Interpreting CPI and SPI Together (Task 3.D)
The two indices read as a quadrant, and each quadrant implies a different management response.
| SPI ≥ 1.0 (ahead / on plan) | SPI < 1.0 (behind plan) | |
|---|---|---|
| CPI ≥ 1.0 | Performing well. Verify EV is not overstated and that AC includes accruals. | Efficient but slow. Usually under-resourced — check whether adding resource is affordable given the favourable CPI. |
| CPI < 1.0 | Fast but expensive. Often acceleration, overtime, or overmanning. Ask what the schedule gain cost. | Both adverse. The serious case: quantify the gap, decompose the cause, and prepare a costed recovery plan. |
Three interpretation rules:
- Check for measurement error before diagnosing performance. An unexpectedly favourable CPI early in a project is far more often missing accruals than genuine efficiency.
- Cumulative, not periodic, drives forecasting. Periodic indices swing on cutoff timing. Cumulative CPI is the forecasting basis, and empirical program data shows it stabilises by roughly 15–20% completion.
- Never read SPI as the schedule. Confirm against the critical path (Section 19.2).
4. Explaining Financial Health to Different Stakeholders (Task 3.B)
Task 3.B is explicit that the audience varies. The underlying analysis does not change; the framing does.
| Stakeholder | The question they are actually asking | What to lead with |
|---|---|---|
| Project sponsor | Will this deliver within the funding I secured? | Forecast at completion versus authorization, and the decision required |
| Finance / treasury | How much cash, in which periods? | The funding profile and its change since last period |
| Operations / end user | When can I have it, and at what running cost? | Forecast handover date and life-cycle cost implications |
| Lender or investor | Is the investment case intact? | Forecast total cost against the case, and the coverage of remaining risk |
| Joint-venture partner | Is my share correctly stated? | Cost share, cash calls, and the audit basis |
| Regulator / public body | Is public money properly controlled? | Governance, change control, and the audit trail |
The three-sentence health statement
A useful discipline is to be able to state financial health in three sentences before writing anything longer:
- Position. "At the June data date the project is 47% complete, with cumulative CPI 0.91 and $6.2M of cost variance."
- Forecast. "Forecast at completion is $148M against a $138M authorization, a $10M overrun, with $3.1M of contingency remaining."
- Action. "A recovery plan re-sequencing mechanical erection recovers an estimated $4M; the board is asked to approve a $6M funding increase or a de-scope of the second train."
If those three sentences cannot be written, the analysis is not finished.
[!IMPORTANT] Never report a forecast without naming its method. "EAC is $148M" is incomplete. "EAC is $148M using BAC divided by cumulative CPI, which assumes current cost performance continues" tells the reader what assumption they are buying. The typical, atypical, and composite methods (Section 12.1) can differ by tens of millions on the same data, and the difference between them is entirely an assumption about the future.
A variance report lists a $340,000 unfavourable cost variance on a mechanical erection account with the cause given as 'lower than planned productivity.' Why is this entry inadequate, and what would make it adequate?
At period 3 of a two-year project, cumulative CPI is 1.22 and SPI is 1.04. The project manager wants to report early cost outperformance to the board. What should the cost engineer check first?
A cost engineer reports: 'Forecast at completion is $148 million against a $138 million authorization.' The sponsor asks whether that figure can be relied on. What is missing from the report?