Use this skill when building budgets, conducting variance analysis, implementing rolling forecasts, or allocating costs. Triggers on FP&A, budgeting, variance analysis, rolling forecasts, cost allocation, headcount planning, department budgets, and any task requiring financial planning or budget management.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
Use this skill when building budgets, conducting variance analysis, implementing rolling forecasts, or allocating costs. Triggers on FP&A, budgeting, variance analysis, rolling forecasts, cost allocation, headcount planning, department budgets, and any task requiring financial planning or budget management.
Budget is a plan, not a constraint - A budget is a hypothesis about the
future. When reality diverges from plan, the job is to understand why and
update the forecast - not to defend the original numbers or cut spending
mechanically to hit a line. A budget that nobody updates is just a document.
Rolling forecasts beat annual budgets - An annual budget is stale the
day it is published. Rolling forecasts (typically 12 or 18 months forward,
updated monthly or quarterly) keep the financial view current with business
reality. Many high-performing FP&A teams use the annual budget for target-
setting and rolling forecasts for operational decision-making.
Variance analysis drives learning - The value of budgeting is not the
budget itself but the discipline of comparing plan to actuals and asking
"why?" Every significant variance is a signal: market changed, assumption
was wrong, execution slipped, or an opportunity emerged. Variance analysis
without root-cause investigation is just arithmetic.
Zero-base periodically - Incremental budgeting ("last year plus 5%")
locks in historical inefficiencies. Zero-based budgeting (ZBB) forces every
dollar to be justified from scratch. ZBB is expensive - do it for a full
business unit every 3-5 years, or for cost categories that have grown faster
than revenue for two consecutive years.
Headcount is the biggest lever - In most knowledge-work businesses,
60-70% of operating expenses are people costs (salaries, benefits, payroll
taxes). Headcount planning is therefore the highest-leverage FP&A activity.
Model headcount at the individual role level, not in aggregate - aggregate
headcount budgets hide timing assumptions and grade-mix shifts.
Variance = Actuals - Budget (favorable if positive for revenue, negative for
expenses; adverse if the opposite). Three decomposition layers:
Volume variance - driven by more or fewer units/transactions than planned
Price/rate variance - driven by a different price or unit cost than planned
Mix variance - driven by a shift in the composition of revenue or cost
Rolling forecasts
A rolling forecast extends the planning horizon by one period every time one
period closes. Instead of a fixed year-end target, the team always looks the
same distance into the future. Cadence options:
Monthly with 12-month horizon - high effort, high accuracy, used by fast-
growth companies where 3-month-old assumptions are obsolete
Quarterly with 4-6 quarter horizon - balanced effort, used by most mid-
market companies
Annual reforecast - minimum viable; update the annual budget once (e.g.
at mid-year) to reflect H1 actuals
Cost centers
Cost centers are organizational units tracked for expense accountability but
not directly linked to revenue. Categorization matters for allocation:
Direct cost centers - produce the product or service (engineering,
manufacturing, customer success delivery)
Indirect cost centers - support the business (HR, finance, IT, legal,
facilities)
Shared services - serve multiple business units and require allocation
Common tasks
Build an annual budget
Use this template sequence to construct a bottom-up operating budget:
Step 1 - Revenue model
Revenue = Volume x Price (by product / segment / channel)
- Prior year actuals as base
- Growth assumptions by segment (market data + sales pipeline + management targets)
- Pricing assumptions (list price, discount rate, mix shifts)
Step 2 - COGS and gross margin
COGS = Variable COGS + Fixed COGS
- Variable: unit costs x volume (hosting, payment processing, direct labor)
- Fixed: depreciation, facilities tied to delivery
Gross Margin % = (Revenue - COGS) / Revenue
Step 3 - Operating expenses by department
For each department:
Headcount costs = (salary + benefits + payroll tax) per FTE x planned FTEs
+ Timing of new hires (partial-year cost for mid-year starts)
Non-headcount = software, contractors, T&E, marketing spend, etc.
Step 4 - EBITDA and cash flow bridge
EBITDA = Gross Profit - OpEx
Cash flow = EBITDA - CapEx - working capital changes - debt service
Step 5 - Scenario analysis
Base case: most likely assumptions
Bear case: 10-20% below base revenue, hold costs at base
Bull case: 15-25% above base revenue, model incremental investment
Conduct variance analysis
Use the FAV/UNF framework to structure every variance report:
For each P&L line:
1. Compute: Actual vs. Budget ($) and (%)
2. Flag: Favorable (FAV) or Unfavorable (UNF)
3. Decompose (if >$X threshold or >5%):
- Is variance volume-driven? (more/fewer units)
- Is variance rate/price-driven? (unit cost or price changed)
- Is it timing? (spend shifted quarters - not a real variance)
- Is it a new item not in budget? (one-time or structural)
4. Root cause: one sentence explaining why
5. Reforecast impact: does this variance repeat in future months?
See references/variance-templates.md for full report formats.
Implement rolling forecasts
Transition from annual budgeting to rolling forecasts:
Lock the current annual budget as the baseline "target" - this is what
compensation and bonuses are measured against
Start a parallel 12-month rolling model updated monthly
In the rolling model, lock the nearest 1-2 months (actuals will replace
them shortly); allow full flexibility in months 3-12
Each month: ingest actuals, roll forward by one month, update assumptions
for months 3-12 based on what changed
Measure forecast accuracy: track MAPE (Mean Absolute Percentage Error)
by line item; target <5% on revenue, <8% on total OpEx
Forecast lock dates (example cadence):
Day 3 after period close: actuals loaded, prior period locked
Day 5: department heads update their forward months
Day 7: FP&A consolidates and runs sanity checks
Day 10: CFO review and final lock
Plan headcount
Build the headcount model at the individual-role level:
For each planned role:
- Job title and grade/level
- Department and cost center
- Start date (month precision)
- Annualized base salary (use midpoint of band)
- Benefits load % (typically 20-30% of base; confirm with HR/payroll)
- Employer payroll taxes (6.2% FICA SS up to wage base + 1.45% Medicare)
- Fully-loaded cost = base x (1 + benefits load + payroll tax %)
- Budget = fully-loaded cost x (months remaining in year / 12)
Contractor model:
- Bill rate x estimated hours (or monthly retainer)
- No benefits load; may carry a premium vs. FTE for flexibility
Track four headcount metrics monthly:
Planned headcount - approved budget positions
Filled headcount - active employees (including notice periods)
Open requisitions - approved but unfilled
Attrition - voluntary and involuntary departures; factor 10-15% annual
attrition into hiring plan to hold steady-state headcount
Allocate costs
Choose the allocation method by cost type:
Method
When to use
Allocation base
Direct
Cost is 100% attributable to one department
N/A - charge directly
Indirect (simple)
Shared cost, easy driver
Headcount, revenue, square footage
Activity-based (ABC)
High shared cost, heterogeneous usage
Actual activity units consumed
Tiered
Large shared service with SLA tiers
Weighted usage by tier
Activity-based cost allocation example:
IT infrastructure cost: $1,200,000/year
Driver: compute units consumed per department (measured from cloud billing)
Engineering: 60% of compute = $720,000
Product: 15% of compute = $180,000
Sales: 10% of compute = $120,000
G&A: 15% of compute = $180,000
Avoid headcount-only allocation for technical shared services - it misprices
costs and subsidizes heavy users at the expense of light users.
Build department budgets
Guide department heads through this template:
Department Budget Template
--------------------------
1. Mission and top priorities for the year (3-5 bullet points)
2. Headcount plan: current FTEs, planned adds, planned attrition, end-of-year
3. Headcount cost (use standard fully-loaded rates from FP&A)
4. Non-headcount detail:
- Software/SaaS subscriptions (list each tool and annual cost)
- Contractors and professional services
- Travel and entertainment (T&E)
- Training and development
- Other (specify)
5. Total department budget
6. Key assumptions and risks
7. Investment requests above baseline (rank-ordered with ROI rationale)
Run budget calibration sessions: compare department requests to company-level
targets; negotiate trade-offs before the final budget is locked.
Present budget to leadership
Structure the budget presentation deck as:
1. Executive summary (1 slide)
- Revenue, gross margin, EBITDA, headcount - plan vs. prior year
- 3 key bets the budget funds
2. Revenue plan (2-3 slides)
- By segment / product / geography
- Growth assumptions and confidence level
- Pipeline coverage ratio
3. Cost structure (2-3 slides)
- Gross margin bridge: prior year to plan
- OpEx waterfall: headcount vs. non-headcount growth
- Cost as % of revenue trend
4. Headcount plan (1-2 slides)
- Net adds by department
- Hiring timing and pipeline status
5. Scenario analysis (1 slide)
- Bear / Base / Bull EBITDA and cash flow
- Key sensitivities (e.g., "$5M revenue miss = $X EBITDA impact")
6. Key risks and mitigations (1 slide)
7. Asks / decisions needed (1 slide)
Lead with the so-what on every slide. CFOs and CEOs do not want to read tables
they want to know what the number means for the business.
Anti-patterns
Mistake
Why it's wrong
What to do instead
Incremental budgeting without review
Locks in historical spend regardless of ROI; 10% growth on wasteful spend is still waste
Zero-base any cost category that grew faster than revenue for 2+ years
Sandbagging revenue / padding costs
Teams build in buffers to hit targets easily; aggregated company plan is materially off
Separate "stretch targets" from "base case" - be explicit about the probability level of each
Monthly budget with no rolling forecast
By Q3, the annual budget is so stale it drives no decisions
Maintain a rolling 12-month forecast alongside the annual budget target
Headcount in aggregate
Hides timing and grade-mix; a "10 headcount" budget might mean very different things
Model every role by title, level, start month, and fully-loaded cost
Allocating shared costs equally by headcount
Misprices costs; a 3-person engineering team using 60% of cloud infrastructure pays the same as a 3-person legal team
Use activity-based drivers that reflect actual consumption
Variance analysis without root cause
"We missed by $200K" is not analysis - it is arithmetic
Every variance above threshold requires a one-sentence root cause and a forward reforecast impact
Gotchas
Partial-year headcount math - A hire starting July 1 costs 6/12 of their annual loaded cost in that fiscal year. Forgetting to prorate new hires is the most common reason department budgets look balanced on paper but come in over on actuals.
Benefits load varies wildly - 20-30% is a guideline but actual loads vary by country, company size, and plan design. Always confirm the exact employer-side benefits load with HR/payroll before locking headcount budgets - using the wrong rate can misstate OpEx by 5-10%.
Rolling forecast ≠ annual budget - Treating the rolling forecast as the new annual budget (and tying comp to it) removes the incentive to forecast accurately. Keep the original annual budget as the compensation target; use the rolling forecast only for operational decisions.
Variance timing traps - A variance that is labeled "timing" (spend shifted quarters) often resurfaces in Q4 as a real overage. Track timing variances separately and flag them if they haven't resolved by month 9.
Allocation method drift - If you switch cost allocation drivers mid-year (e.g., from headcount to compute units), prior-period comparisons become meaningless. Lock the allocation method at budget time and don't change it within a fiscal year without a full restatement.
References
references/variance-templates.md - Variance analysis templates and
reporting formats for monthly, quarterly, and ad-hoc variance reports
Only load a reference file if the current task requires the detailed formats
or templates within it.
Companion check
On first activation of this skill in a conversation: check which companion skills are installed by running ls ~/.claude/skills/ ~/.agent/skills/ ~/.agents/skills/ .claude/skills/ .agent/skills/ .agents/skills/ 2>/dev/null. Compare the results against the recommended_skills field in this file's frontmatter. For any that are missing, mention them once and offer to install:
