Purpose
Take budget-versus-actual and forecast-at-completion data for a capital project and produce a structured root cause analysis that attributes the total overrun to specific, weighted drivers — so leadership knows exactly where cost leaked, how much each cause contributed, and what to fix on this project and the next one. Replaces the vague "we had scope changes and material inflation" narrative with a quantified, defensible attribution.
When to Use
Invoke this skill when the user asks to:
- Investigate why a capital project is over budget or trending over its approved baseline
- Attribute a forecast-at-completion (EAC) overrun to specific root causes
- Prepare a cost variance root cause narrative for a steering committee, board, or CFO
- Diagnose whether an overrun is an estimating problem, an execution problem, or a market problem
- Break down a lump-sum variance into weighted contributing factors
- Build lessons-learned findings from a cost overrun for future project governance
Inputs Required
Before starting, collect:
- Project description: Type, scope, delivery method (e.g., EPC, design-bid-build, CM-at-risk), and current phase
- Approved baseline budget: Original sanctioned/control budget and any approved change orders to date
- Actuals to date: Committed and incurred cost by major cost category or WBS element
- Forecast at completion (EAC): Current EAC and the resulting variance at completion vs. baseline
- Variance explanations on record: Any existing project team commentary on cost drivers
- Schedule status: Current vs. baseline completion date, and whether the project is behind
- Change log summary: Count and dollar value of change orders, split by scope-add vs. rework/error if available
If a category breakdown or change log isn't available, work from the total variance and flag which attributions are estimates requiring team validation.
Instructions
Step 1: Establish the Variance Baseline
Quantify what is actually being explained before assigning cause:
- State the total variance at completion = EAC − (approved baseline + approved change orders), in dollars and percent
- Separate committed/incurred variance (already spent over) from forecast variance (projected to spend over) — these have different urgency and recoverability
- Note the schedule variance if present — cost overruns and schedule slips are frequently the same event counted twice
- Establish the denominator all weighting will resolve to: the total variance is 100%
Step 2: Diagnose Across Five Root Cause Categories
Assess each category, assign an estimated dollar contribution and confidence level, and cite the evidence.
Category 1 — Scope Growth Cost added by work that was not in the sanctioned baseline scope.
- Look for: net-additive change orders, stakeholder-driven additions, gold-plating, uncontrolled requirements creep, scope not baselined at sanction
- Distinguish authorized scope growth (approved additions — legitimate but signals baseline/estimating maturity issues) from scope creep (crept in without formal change control — a governance failure)
- Evidence signals: change order log dominated by "add" line items; growth in quantities vs. estimate takeoff; late stakeholder requirements
Category 2 — Estimating Error Cost the baseline should have captured but didn't.
- Look for: missing scope in the original estimate, quantity takeoff errors, inadequate contingency for the class of estimate, optimistic productivity or unit-rate assumptions, escalation not modeled
- Check the estimate class — a Class 4/5 estimate carries ±30–50% expected accuracy; overruns within that band are estimate immaturity, not execution failure
- Evidence signals: overruns concentrated in categories with no change orders (i.e., the budget was simply wrong); contingency drawn down early; unit rates below market at sanction
Category 3 — Productivity Loss (Execution) Cost from performing the planned scope less efficiently than estimated.
- Look for: labor hours exceeding budgeted hours at the same physical progress, rework/quality failures, out-of-sequence work, congestion/stacking of trades, weather and access disruption, poor field supervision
- Use earned-value logic where possible: CPI = BCWP/ACWP below 1.0 on installed quantities isolates execution inefficiency from scope and estimate causes
- Evidence signals: hours variance without quantity variance; rework charges; disruption/acceleration claims
Category 4 — Supply Chain & Escalation Cost from market and procurement conditions outside the estimate assumptions.
- Look for: material price escalation above assumed indices, labor market tightness/wage inflation, currency movement, expediting/premium freight, single-source dependency, vendor delivery failures forcing substitutions
- Separate escalation (price movement over time) from procurement execution (poor bid strategy, late buyouts losing pricing) — the first is market, the second is controllable
- Evidence signals: buyout values above estimate for unchanged specs; documented index movement; premium logistics charges
Category 5 — Design Change & Development Cost from design evolution, errors, and omissions.
- Look for: design errors and omissions driving rework, late design changes, constructability issues, incomplete design at award (IFC gaps), regulatory/permit-driven redesign
- Distinguish design error/omission (contractor or engineer error — potentially recoverable/insurable) from design development (expected maturation from concept to detailed design — an estimating maturity issue)
- Evidence signals: E&O change orders, RFI volume above norm, revisions to issued-for-construction documents
Step 3: Weight the Contributing Factors
Resolve all category contributions to 100% of the total variance:
- Assign each of the five categories a dollar contribution and a percent of total variance
- Where causes overlap (e.g., a design change that also caused productivity loss), attribute the primary driver and note the secondary effect — do not double-count
- Assign a confidence level (High / Medium / Low) to each attribution based on evidence quality; Low-confidence attributions must be flagged for team validation
- Identify the single largest root cause and whether the overrun is fundamentally a pre-sanction problem (estimating, design maturity) or an execution problem (productivity, procurement, scope control)
Step 4: Systemic vs. Project-Specific Determination
Classify each material root cause as:
- Project-specific — a one-off circumstance unlikely to recur
- Systemic — a pattern that will repeat on other projects unless governance changes (e.g., chronic under-contingency, weak change control, immature estimates at sanction) This determines whether the fix is a project recovery action or a PMO/governance change.
Step 5: Corrective & Recovery Actions
Produce two action sets:
- Recovery actions (this project): What can still be controlled on remaining forecast cost — ranked by dollar impact (e.g., freeze scope, recover E&O claims, resequence to reduce disruption, renegotiate open buyouts)
- Governance actions (future projects): What must change for systemic causes (e.g., raise contingency for this estimate class, enforce change control gate, require Class 3 estimate before sanction)
Output Format
- Structure: Executive attribution summary first, then category-by-category analysis, then weighting table, then actions
- Length: 900–1,300 words
- Format: Structured document with a weighting table; category sections as labeled analysis blocks; actions as ranked lists
- Sections: Variance Baseline → Root Cause Analysis (5 categories) → Contributing Factor Weighting Table → Systemic vs. Project-Specific → Recovery Actions → Governance Actions → Single Largest Root Cause verdict
- Tone: Project controls director — evidence-led, quantified, plain about what is and isn't defensible; never speculative without labeling it as such
Quality Checklist
Before delivering output, verify:
- [ ] The category weightings sum to 100% of the stated total variance (check the math)
- [ ] Every category has a dollar contribution AND a confidence level — no category left unassessed
- [ ] Authorized scope growth is separated from uncontrolled scope creep, and escalation is separated from procurement execution
- [ ] Each attribution cites specific evidence signals — not "likely scope issues" with no basis
- [ ] A single largest root cause is named and classified as pre-sanction vs. execution
- [ ] Recovery actions are ranked by dollar impact and address only controllable remaining forecast
Notes
- The most common misattribution is calling an estimating error a "scope change" — if there is no change order but the category is over, the budget was wrong, not the scope
- Cost overruns and schedule slips are frequently the same root event; when a project is both over budget and late, check whether productivity loss or disruption is being double-counted across the two variances
- Contingency drawdown is the earliest warning sign — a project that consumed its contingency before 50% completion almost always has a pre-sanction estimating or design-maturity root cause
- When evidence is thin, deliver the attribution as a hypothesis set with confidence levels and a validation checklist for the project team, rather than presenting low-confidence numbers as fact
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills