The Monthly Analysis Rhythm
Variance analysis is not a one-time event — it is a monthly discipline. Each month, after the books close and EV is calculated, the CAM must review performance data, identify significant variances, determine root causes, and either confirm existing corrective actions or develop new ones. The rhythm should feel routine, not reactive.
Validate the Data
Before analyzing variances, verify the inputs. Are actuals correct? Were time charges posted to the right accounts? Is the earned value claim accurate? Garbage in, garbage out. Spend 15 minutes validating before spending 2 hours analyzing.
Calculate Current and Cumulative Variances
Cost Variance (CV) = BCWP – ACWP. Schedule Variance (SV) = BCWP – BCWS. Calculate both current-period and cumulative. A current-period variance shows what happened this month. A cumulative variance shows the trend. Both matter.
Segregate by Cause
Do not report a lump-sum variance. Break it down: how much is rate vs. usage? How much is labor vs. material? How much is schedule-driven vs. efficiency-driven? The segregation reveals the root cause and points to the right corrective action.
Write the Narrative and Corrective Action
Document the variance cause, impact, and corrective action. Be specific. Include dates, responsible parties, and measurable milestones. Generic corrective actions like “we will work harder” are not corrective actions.
Segregating Variances by Cause
A cost variance of –$50K tells you almost nothing. Is it because you paid people more than planned? Used more hours than planned? Had unplanned material costs? Each cause demands a different corrective action. Segregation is how you move from “we have a problem” to “we understand the problem.”
| Variance Type | Formula | What It Reveals | Example |
|---|---|---|---|
| Rate Variance | (Budgeted Rate – Actual Rate) × Actual Hours | Are you paying more or less per hour than planned? | Planned $80/hr, actual $95/hr on 500 hours = –$7,500 rate variance |
| Usage (Efficiency) Variance | (Budgeted Hours – Actual Hours) × Budgeted Rate | Are you using more or fewer hours than planned for the work accomplished? | Planned 400 hrs, actual 500 hrs at $80/hr = –$8,000 usage variance |
| Schedule Variance (hours) | BCWP – BCWS (in hours or dollars) | Is work being accomplished on time? | Planned $40K of work, earned $30K = –$10K schedule variance |
| Mix Variance | Variance caused by using different labor categories than planned | Are you using the right skill levels? | Planned junior engineers at $70/hr, used senior engineers at $110/hr |
💡 The Rate vs. Usage Distinction Matters
A rate variance is often outside the CAM’s control — it is driven by labor rate changes, overhead adjustments, or using a different labor mix than planned. A usage variance is squarely in the CAM’s domain — it means the work took more or fewer hours than budgeted. Do not let a rate variance hide a usage problem or vice versa. Segregate them so you can act on what you can control and escalate what you cannot.
Writing Effective Variance Narratives
The variance narrative is the CAM’s primary communication tool to management and the customer. A well-written narrative builds confidence that the CAM understands the work and has a plan. A poorly written one raises questions and invites deeper scrutiny.
❌ Poor Narrative
- “Cost variance is unfavorable due to higher than planned costs.”
- “Schedule variance caused by late deliveries.”
- “Corrective action: monitoring closely.”
- “Expect performance to improve next month.”
- No quantification of causes
- No specific dates or milestones for recovery
✅ Strong Narrative
- “CV of –$47K driven by: (1) –$30K usage variance from 375 unplanned rework hours on thermal test Unit 3; (2) –$17K rate variance from Sr. Engineer backfill at $110/hr vs. planned $85/hr.”
- “Root cause: design tolerance inadequacy identified during TRR, requiring test article rework.”
- “Impact to EAC: +$55K (rework complete, no further impact expected).”
- “Corrective actions: (1) ECR-0042 design revision — completion 15 Mar. (2) Backfill position filled with mid-level engineer effective 1 Apr, eliminating rate variance.”
The four-part structure that every narrative should follow:
| Section | Content | Key Test |
|---|---|---|
| What happened? | Quantified variance by cause category. Break into rate, usage, mix, material, subcontract. | Do the pieces sum to the total variance? |
| Why did it happen? | Root cause, not restatement of the numbers. Technical cause, not financial symptoms. | Would someone unfamiliar with the project understand the cause? |
| What is the impact? | Effect on EAC, schedule milestones, other CAs, or customer deliverables. | Is the impact quantified in dollars and/or schedule days? |
| What are you doing about it? | Specific corrective actions with owners, dates, and measurable milestones. | Could an auditor verify whether the corrective action was completed? |
Corrective Actions with Measurable Milestones
A corrective action without a measurable milestone is a hope, not a plan. Every corrective action should specify what will be done, who is responsible, when it will be complete, and how you will verify it worked.
| Weak Corrective Action | Strong Corrective Action |
|---|---|
| “Will work overtime to recover schedule.” | “Authorized 10 hrs/week OT for 3 engineers through 30 Apr. Target: recover Milestone 4 by 15 May (original date 1 May). Cost impact: +$12K factored into EAC.” |
| “Monitoring the situation closely.” | “Weekly review of rework hours vs. plan initiated. If rework exceeds 200 additional hours by 15 Mar, will escalate to PM for MR request.” |
| “Expect improvement next quarter.” | “New test procedure (TP-0088) eliminates redundant thermal cycle, saving 120 hrs. Procedure effective 1 Apr. Expected improvement: CPI from 0.87 to 0.93 by end of Q3.” |
Variance Trends and Early Warning
The most valuable variance analysis looks beyond the current month. Trends reveal systemic problems that single-month data cannot. A CPI that drops from 1.02 to 0.98 to 0.95 over three months is a signal even if no single month exceeds the threshold.
💡 The Cumulative CPI Rule
Cumulative CPI is remarkably stable after 20% of the work is complete. Research across hundreds of DoD programs shows that cumulative CPI rarely improves by more than 10% from the 20% completion point. If your cumulative CPI is 0.85 at 20% complete, history says it will finish between 0.83 and 0.93. This means early variances matter more than late ones — they are not anomalies, they are forecasts. Do not dismiss early negative trends with “it will get better.” It almost never does.
🎯 The Bottom Line
Variance analysis is where the CAM demonstrates command of their control account. Segregate variances by cause to understand what is really happening. Write narratives that answer the four questions: what, why, impact, and corrective action. Make corrective actions specific with measurable milestones. Watch trends, not just thresholds. The monthly analysis discipline is what separates a CAM who manages from one who merely reports. Next: Estimate to Complete — how to translate variance analysis into a credible forecast of future costs.
The corridor is the page's own rule: across hundreds of DoD programs, cumulative CPI rarely improves by more than 10% from the 20% completion point, so 0.85 at the gate finishes between 0.83 and 0.93. It is drawn as a wedge opening from one point rather than as two rails, because the rule is about the finishing value and says nothing about the path between; parallel rails would assert month-by-month bounds the research does not give. Nothing is drawn before 20% either — the page says the number is not stable there, and a line through it would invent the one thing the rule refuses to predict. The dashed line is the recovery a red control account is normally replanned to deliver. It leaves the corridor within a few per cent of the gate.
Interactive Demo
The same overrun, segregated into rate and usage. Defaults are the page's own worked figures — then try pushing the rate up and the hours down until the total reads clean and both causes are still there.
Stop reading, start modeling
Model your process flow, run simulations, optimize staffing with TOC math, and test your knowledge with 107 interactive checks — all in one platform.