What is the difference between a CAR and a VAR?

A Corrective Action Report (CAR) is generated when a control account breaches a predefined threshold and requires a formal corrective action plan with milestones. A Variance Analysis Report (VAR) is the monthly narrative explaining WHY a variance exists. In practice, many organizations combine them: the VAR explains the root cause, and when thresholds are breached, a CAR section is appended with specific corrective actions, responsible parties, and target dates.

How do I determine the right variance thresholds?

Common thresholds are ±10% AND ±$100K at the control account level — both conditions must be met to trigger a formal report. Some contracts specify thresholds in the CDRL or the EVMS System Description. Thresholds should be tiered: stricter for large control accounts, looser for small ones. A $5M CA with a 10% threshold triggers at $500K; a $50K CA at the same percentage triggers at only $5K, which may be noise.

What makes a good variance narrative?

A good narrative answers five questions: (1) What is the variance? (2) Why did it occur — the root cause, not the symptom? (3) What is the impact to the program? (4) What corrective action is being taken? (5) When will performance return to plan? Bad narratives say 'we are over budget due to higher costs.' Good narratives say 'labor rates exceeded plan by 12% because two senior engineers replaced the planned mid-level staff, adding $85K this period.'

Variance Analysis & Reporting: Root Cause Narratives That Drive Action

New to this topic?

We recommend reading these guides first to get the most out of this one:

CPI & SPI Analysis

±10%

Typical CV/SV Threshold

Root Cause Questions

Variance Components

30 days

CAR Response Window

Why Most Variance Narratives Fail

Every month, Control Account Managers across thousands of defense programs write variance narratives. The vast majority are useless. They restate the numbers in words — “The control account is $150K unfavorable due to higher-than-planned costs” — and call it analysis. That is not analysis. That is reading a spreadsheet aloud.

Effective variance analysis answers a fundamentally different question: why did performance deviate from plan, and what are we going to do about it? The narrative must trace the variance to a root cause, quantify its impact, and propose corrective action with a timeline. Anything less is administrative theater.

The monthly variance analysis rhythm is the heartbeat of EVMS. When done well, it provides program managers with the insight they need to make decisions before small problems become large ones. When done poorly, it generates paperwork that nobody reads.

CAR/VAR Thresholds and When They Trigger

Variance thresholds define when a control account requires formal written analysis. Most contracts specify thresholds in the Contract Data Requirements List (CDRL) or the program’s EVMS System Description. The standard approach uses a dual-condition test: both a percentage threshold and a dollar threshold must be breached.

Threshold Type	Typical Value	Applies To	Trigger Logic
Cost Variance (CV)	±10% AND ±$100K	Control Account cumulative	Both conditions must be met
Schedule Variance (SV)	±10% AND ±$100K	Control Account cumulative	Both conditions must be met
VAC Threshold	±10% of CA BAC	Variance at Completion	Single condition
Program-Level	±5% of PMB	Total program CV or SV	Triggers escalation to customer

When a control account breaches its threshold, the CAM must produce a Variance Analysis Report (VAR). If the variance is significant or persistent, the program may escalate to a Corrective Action Report (CAR), which adds formal commitments: specific actions, responsible parties, target completion dates, and follow-up tracking.

💡 Threshold Tiering Prevents Noise

A flat $100K threshold means a $50K control account never triggers formal analysis — even at 200% overrun — while a $10M control account triggers at just 1% deviation. Smart programs tier their thresholds: small CAs use tighter dollar thresholds ($25K), large CAs use looser ones ($250K). The goal is to capture meaningful variances without drowning CAMs in paperwork for rounding errors.

Segregating Variances: Rate, Usage, and Schedule

A cost variance of –$200K tells you almost nothing by itself. To understand the root cause, you must decompose it into its components. Every cost variance can be segregated into three categories:

Component	Definition	Formula	Example Root Cause
Rate Variance	Actual rates differ from planned rates	(Actual Rate – Planned Rate) × Actual Quantity	Senior engineers assigned instead of mid-level; overtime premium; subcontractor rate increase
Usage Variance	More or fewer resources used than planned	(Actual Quantity – Planned Quantity) × Planned Rate	Rework required; scope growth; efficiency gain; learning curve
Schedule Variance	Work performed ahead of or behind the plan	EV – PV (time-based)	Late material delivery; resource unavailability; predecessor slip

📊 Decomposing a $200K Cost Variance Worked Example

Scenario: Control Account 1.3.2 (Software Integration) shows CV = –$200K cumulative.

Element	Planned	Actual	Delta
Labor Hours	2,000 hrs	2,400 hrs	+400 hrs (usage)
Labor Rate	$75/hr	$85/hr	+$10/hr (rate)
Planned Cost (PV)	$150,000	—	—
Actual Cost (AC)	—	$204,000	—

Rate Variance: ($85 – $75) × 2,400 hrs = –$24,000 (10/hr premium on senior staff)

Usage Variance: (2,400 – 2,000) × $75 = –$30,000 (400 extra hours for rework)

Combined Explanation: $24K from rate (senior engineers substituted for mid-level) + $30K from usage (integration rework after interface change). The remaining variance traces to prior periods. Now management knows: the rate issue requires a staffing plan correction; the usage issue requires an interface control process fix.

Writing Narratives That Drive Action

Every variance narrative should follow a five-part structure. This is not optional — it is the difference between a narrative that gets read and one that gets filed.

Section	Question Answered	Bad Example	Good Example
1. State the Variance	What is the number?	“CV is unfavorable.”	“CV = –$200K (–13.3%) cumulative, –$45K current period.”
2. Root Cause	Why did it happen?	“Higher than planned costs.”	“Two SR engineers replaced planned MID staff due to skill gap on radar interface; 400 hrs of rework from ICN-042 interface change.”
3. Impact	What does it mean?	“May affect EAC.”	“Drives EAC increase of $180K; delays CDR entry by 2 weeks.”
4. Corrective Action	What are we doing?	“Will monitor closely.”	“(a) Onboarding 2 MID engineers by 15 Mar; (b) Baselined interface freeze after ICN-042 closure.”
5. Recovery Timeline	When will it improve?	“Expect improvement soon.”	“Current-period CV will return to ±5% by Apr; cumulative CV will stabilize at –$200K.”

Notice the pattern: bad narratives use vague language and restate what the numbers already show. Good narratives name specific causes, quantify specific impacts, and commit to specific actions with dates. The program manager reading this narrative can make a decision. The program manager reading the bad version cannot.

The Monthly Variance Analysis Rhythm

Variance analysis is not a one-time event — it is a monthly discipline that follows a predictable cadence. The rhythm ensures that data flows from the accounting system through analysis to management action within a compressed timeline.

Day	Activity	Owner
Day 1–3	Monthly cost close: actuals posted, EV claimed, schedule statused	Finance / Scheduler / CAMs
Day 4–5	Data validation: reconcile AC, EV, and schedule; resolve discrepancies	EVMS Analyst
Day 6–8	CAMs write variance narratives for breached thresholds	CAMs
Day 9–10	Program-level review: roll up narratives, identify systemic issues	PM / Program Controls
Day 11–15	CPR/IPMR assembly and submission to customer	Program Controls

The critical path runs through data quality. If actuals are not posted correctly by Day 3, every downstream activity slips. Programs that struggle with variance analysis almost always have an upstream data problem — not a narrative-writing problem.

⚠️ The “Will Monitor” Anti-Pattern

If your corrective action plan says “will continue to monitor,” you do not have a corrective action plan. Monitoring is not action — it is the absence of action. Every CAR must include at least one concrete step: reassign staff, reschedule work, descope, change the technical approach, or request a baseline change. If none of these apply, then either the variance is not significant enough to warrant a CAR, or the team has not yet identified the real problem.

🎯 The Bottom Line

Variance analysis is not about explaining numbers — it is about identifying root causes and driving corrective action. Segregate every variance into rate, usage, and schedule components to find the real problem. Write narratives that answer five questions: what, why, impact, action, and timeline. Follow the monthly rhythm religiously, and never accept “will monitor” as a corrective action. Next: Estimate at Completion Methods — translating today’s performance into tomorrow’s forecast.

Keep Learning

📖

Guide

Estimate at Completion Methods

📖

Guide

Monthly CPR/IPMR Generation

🏭

Free Process Modeler

Map your production flow, find bottlenecks & optimize staffing. No login required.

Try It Free →

💾