# Usage: curl -sSL https://seed.show/finance.planning.analysis | bash -s <install-path>
# <install-path> is the directory where the file should land.

set -euo pipefail
[ -z "${1:-}" ] && {
  echo "install requires a path: curl -sSL https://seed.show/finance.planning.analysis | bash -s <install-path>" >&2
  exit 1
}
TARGET="$1"
mkdir -p "$TARGET"
DEST="$TARGET/seed-fold.eNOYpS.folded.md"

cat > "$DEST" <<'PORTDOWN_EAECF7B5'
<!--
This is a `.folded.md` archive — a directory packed into one markdown
file. The markers below are load-bearing; don't edit them directly.

To unpack (requires bash — if you have no shell, tell the user):
  1. curl -sSLf https://fold.dom.vin/skill | bash -s <INSTALL_DIR>
  2. <INSTALL_DIR>/fold/scripts/unfold <this-file>
     (or: unfold <this-file>  if fold/scripts is on your PATH)
-->

---
fold: true
marker: 432768
at: 2026-05-07T16:16:49Z
root: seed-pack.2lcKAP
---

<!--fold:432768@file path="README.md" mode="644"-->
# Financial Planning & Analysis (FP&A) Context

## What this is and what it is not

This context is for agents assisting FP&A practitioners: analysts, finance business partners, controllers, and CFOs working through planning, forecasting, and variance analysis tasks.

**Do not generate company-specific financial figures.** Do not fabricate revenue numbers, margin percentages, or headcount counts — these must come from the user's actual data. Do not assume any particular company's financial structure, planning calendar, or reporting hierarchy. What this context provides is structural FP&A logic: how planning cycles work, how variance analysis is framed, how driver-based models are constructed, what distinguishes a budget from a forecast.

---

## Mental model: FP&A as the translation layer

FP&A sits between strategy and operations. Its job is to translate management's strategic intent into a financial plan, then continuously compare that plan against reality and surface the gap in terms that inform decisions.

The model is a decision tool, not a prediction. A budget is not a forecast of what will happen; it is management's theory of what they intend to make happen. A rolling forecast is not an update to the budget; it is a continuously refreshed view of what now looks likely. The distinction matters because they are used for different things: budgets for accountability, forecasts for decisions.

FP&A's value is not in the spreadsheet — it is in the translation. Any analyst can sum a column. The work is knowing which numbers to look at, what explains the gap between plan and actual, and what management needs to know before the next decision.

---

## The planning cycle

### Budget (annual plan)
- Set once per year, typically Q4 for the following fiscal year
- Approved by leadership and board; serves as the official performance target
- Fixed for the year — variances are measured against it, not corrected by revising it
- Built bottom-up (managers submit plans) or top-down (leadership sets targets, teams backfill), usually both in negotiation
- The budget is a commitment, not a prediction

### Rolling forecast
- Updated continuously, typically monthly or quarterly
- Always covers a fixed horizon forward (e.g., next 12 months, or current FY remaining + 1 quarter)
- Incorporates current actuals and latest business signals
- Not a budget revision — it coexists with the fixed budget as a separate, forward-looking view
- Better for operational decision-making; worse for accountability (it can be adjusted, so it can't be held against)

### Scenario plan
- Answers "what if" — not what we expect, but what would happen under specific alternative conditions
- Scenarios are named and bounded: "downside: revenue −20%, hiring freeze"; "base: plan"; "upside: accelerated expansion"
- Used for stress-testing, board communication, and contingency planning
- Scenarios are not probability-weighted; they are discrete narratives with full P&L and cash implications
- Distinct from sensitivity analysis, which varies a single assumption while holding others constant

---

## Variance analysis framework

Variance analysis is the core analytical cycle in FP&A. Every month, FP&A compares actuals to budget (BvA) and to the prior forecast, explains the gaps in terms management can act on, and updates the forward view.

Variance is always decomposed, not just reported. The decomposition categories:

**Volume variance:** The impact of doing more or less of something than planned — more units sold, more hires, more projects. Volume variances are often operationally explainable and partially recoverable.

**Price/rate variance:** The impact of a different price, rate, or cost per unit than assumed — higher average selling price, lower vendor costs, higher-than-planned salary rates. Price variances often reflect market or negotiating conditions.

**Mix variance:** The impact of a different composition of activity than planned — more revenue from lower-margin products, more headcount in higher-cost geographies. Mix variances are frequently invisible until isolated.

**Timing variance:** The impact of activity happening in a different period than planned — a deal that slipped to the next quarter, a hire that started late. Timing variances are often recoverable; they are not losses, they are displacements.

Strong variance commentary isolates which of these is at work, segments the impact (which product, segment, or geography), and states the forward-looking implication. Weak commentary names the number and moves on.

---

## What agents get wrong

### 1. Conflating budget with forecast
The budget and the rolling forecast are different documents serving different purposes. The budget does not change when the business changes; the forecast does. Telling management "your forecast is above budget" is informative. Saying "let's update the budget to reflect current trends" is a red flag — budgets are not revised mid-year except under extraordinary circumstances (acquisition, restructuring, board-approved reset).

### 2. Point estimates without ranges
FP&A professionals almost never present a single-number forecast without acknowledging uncertainty. Real forecasts carry ranges, confidence intervals, or scenario brackets. An agent that produces a single revenue figure for Q3 without a range or variance assumption is presenting false precision. The appropriate response to uncertainty is to show the distribution, not hide it in a midpoint.

### 3. Variance commentary that describes rather than explains
Weak: "Revenue was $2.1M versus budget of $2.3M, a $200K unfavorable variance."

Strong: "Revenue came in $200K below budget. The gap is driven by two deals that slipped to Q4 (combined $310K), partially offset by a $110K upside in existing customer expansion. Deal slippage was concentrated in the enterprise segment; SMB performed in line with plan. Both slipped deals are expected to close in October."

The weak version is arithmetic. The strong version decomposes the variance by type (timing), segments it (enterprise vs. SMB), and provides a forward view. Agents default to the weak version because it requires only the numbers; the strong version requires understanding the business context behind them.

### 4. Treating driver-based models as black boxes
FP&A models are built on assumptions about business drivers — headcount drives payroll, ARR drives revenue, traffic drives conversion drives pipeline. An agent that replicates formulas without understanding the underlying driver logic will produce forecasts that are arithmetically correct but operationally meaningless. The first question for any line item: what are the 3–5 key drivers, and where do the driver assumptions come from?

### 5. Ignoring the cash vs. P&L distinction
Revenue on the P&L is not cash in the bank. Deferred revenue, unbilled AR, and capex vs. opex distinctions matter enormously in FP&A. A SaaS company can show strong ARR growth while burning cash. FP&A produces three linked views — P&L, balance sheet, and cash flow statement — and the story they tell can diverge significantly. Agents that work only from the income statement are missing at minimum a third of the picture.

### 6. Applying trend extrapolation where driver logic is required
Fitting a trendline to historical revenue and calling it a forecast is not FP&A — it is statistics. Driver-based models are more defensible because they make the causal assumptions explicit. When a sales rep ramp curve, a pricing assumption, or a conversion rate changes, the driver model propagates the effect coherently; the trendline just wiggles.

---

## What AI is changing

### Where AI is being applied now
- **Automated variance narrative:** LLMs generating first-draft variance commentary from structured data — mapping line-item deltas to templated explanations, flagging anomalies for analyst review. Saves time on mechanical description; still requires analyst judgment to determine cause and forward implication.
- **Rolling forecast automation:** ML models trained on historical actuals plus leading indicators (pipeline stage, web traffic, hiring activity) to generate probabilistic revenue and cost forecasts. Reduces cycle time; shifts analyst work toward assumption review rather than model construction.
- **Scenario modeling at scale:** AI-assisted scenario generation — given a macro assumption change (e.g., interest rates +150bps), produce the downstream P&L and cash flow implications. Useful for board pre-reads and stress tests. Risk: plausible-looking outputs that embed flawed causal assumptions.
- **NLP for management commentary:** Generating board-quality prose from structured financial summaries — pulling from prior period narratives, variance tables, and executive KPIs to produce a first draft. Requires human review for strategic framing and tone.
- **Anomaly detection:** Flagging unusual line-item movements in the general ledger before close — catching coding errors, unusual accruals, or unexpected cost spikes in the data before they surface in the BvA.

### What stays human
- **Assumption setting:** The forecast is only as good as its assumptions. Deciding what growth rate, churn rate, or hiring pace to assume is a judgment call that requires understanding the business, the competitive environment, and management's intent. AI can surface historical ranges; it cannot make the call.
- **Strategic framing:** Translating a financial result into a strategic implication — "we're underperforming on enterprise because we haven't closed the product gap, not because we have a pipeline problem" — requires contextual knowledge AI doesn't have access to in the absence of a very rich prompt.
- **Board and executive communication:** The CFO's job in the board room is to tell a coherent story under pressure, field questions about assumptions, and make real-time judgment calls about what to disclose and how. No current AI system reliably does this.
- **Judgment on outliers:** Deciding whether an anomalous result is a data error, a one-time event, or the leading edge of a trend requires domain knowledge and pattern recognition that goes beyond what is visible in the numbers. Analysts are hired for this; AI is not yet there.
<!--fold:432768@file path="glossary.md" mode="644"-->
# FP&A Glossary

Precise definitions for terms frequently misused or conflated in AI-generated FP&A content.

---

**Actuals**
Recorded financial results for a completed period — revenue, expenses, headcount, and other metrics as they actually occurred, per the general ledger. Actuals are backward-looking and fixed once the period closes. The primary input to variance analysis.

**Bottom-Up Forecasting**
An approach where individual managers or business unit leaders build their own plans based on known pipeline, headcount plans, and project commitments. FP&A aggregates them into a consolidated view. More operationally grounded; can understate ambition if managers sandbag or lack strategic context. In practice, most organizations use a hybrid: leadership sets a target, teams build bottom-up, and the gap is negotiated.

**Bridge Analysis (Waterfall Analysis)**
A visual and analytical technique that decomposes the difference between two financial figures — typically prior period to current period, or budget to actuals — into component causes. Each bar in the waterfall represents a contributing factor (volume, price, mix, timing) that "walks" from the starting value to the ending value. Useful for explaining variance to leadership in a single chart: "We went from $10M to $8M. Here's the $3M unfavorable volume effect, the $1M favorable mix effect, and the $2M favorable pricing effect." The most common chart type in FP&A board reporting.

**Budget (Annual Plan)**
A fixed financial plan for the fiscal year, approved by leadership and the board before the year begins. Represents management's commitment to a performance target. Not updated mid-year in normal operations. All variance is measured against the budget as the baseline.

**BvA (Budget vs. Actuals)**
The core monthly reporting exercise in FP&A: comparing what was budgeted for a period to what actually happened. The output is a variance report with favorable (F) or unfavorable (U) flags on each line item, plus explanatory commentary. Also called Budget vs. Actual or Budget vs. Plan.

**Driver Tree**
A visual decomposition of a financial outcome into its causal drivers, arranged hierarchically. Example: Revenue → [New ARR + Expansion ARR − Churn ARR]; New ARR → [Deals Closed × Average Contract Value]; Deals Closed → [Leads × Conversion Rate × Close Rate]. A driver tree makes the model's causal structure explicit, reveals where assumptions live, and shows how a change in one input propagates through to the output. Essential for auditing a driver-based model and for communicating model logic to non-FP&A stakeholders.

**Driver-Based Model**
A forecasting model built on explicit business drivers — the key inputs that causally determine financial outputs. Examples: headcount drives payroll; new logo ARR drives revenue; web traffic × conversion rate × deal size drives pipeline. Driver-based models are more auditable and defensible than trend-extrapolation models; changing a driver assumption propagates through the model coherently. The alternative — fitting a trendline to history — is faster but makes no causal claims and breaks immediately when the business changes.

**EBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization)**
A non-GAAP profitability metric widely used in FP&A as a proxy for operating cash generation. Strips out financing structure (interest), tax position, and non-cash charges (D&A). Commonly used in valuation multiples and covenant calculations. EBITDA is not cash flow — it ignores working capital changes and capex. The gap between EBITDA and free cash flow can be large, particularly in capital-intensive or high-growth businesses.

**Free Cash Flow (FCF)**
Operating cash flow minus capital expenditures. The cash a business generates after maintaining and investing in its asset base. FP&A uses FCF as a fundamental measure of financial health — profitable companies can still destroy value if they consume more cash than they generate. FCF forecasts are derived from P&L and balance sheet models together; agents working only from the income statement cannot produce a valid FCF forecast.

**Headcount Planning**
The FP&A sub-process of forecasting employee count and associated costs (salary, benefits, employer taxes, equity) by role, department, and hire date. Often the single largest expense line for technology and professional services companies. Headcount plans are dynamic: open requisitions, backfills, and attrition all affect the cost forecast before period close. A headcount plan is not a hiring plan — it is a financial model of the workforce, updated as actual hiring and attrition occur.

**Monte Carlo Simulation**
A probabilistic modeling technique that runs a financial model thousands of times, each time drawing driver assumptions from specified probability distributions, to produce a distribution of outcomes rather than a single point estimate. Used in FP&A for cash flow forecasting, capex planning, and scenario stress-testing. The output is a probability-weighted range — "there is a 75% probability that EBITDA lands between $X and $Y" — rather than a deterministic forecast. Agents that produce single-point forecasts without ranges are implicitly ignoring this entire analytical dimension.

**Opex vs. Capex**
Operating expenditures (Opex) are expensed in the period incurred — salaries, rent, software subscriptions. Capital expenditures (Capex) are investments in long-lived assets that are depreciated over time — equipment, leasehold improvements, internally developed software meeting capitalization criteria. The distinction affects both the P&L (timing of expense recognition) and cash flow (capex is a cash outflow immediately, but P&L sees only the depreciation). FP&A models both separately. Misclassifying a significant capex item as opex overstates expenses in the current period and understates the asset base.

**Reforecast**
A mid-year update to the financial forecast that replaces the prior forecast view for the remainder of the fiscal year. Distinguished from a rolling forecast (which is always maintained) and from a budget revision (which is rare and formal). Reforecasts are typically triggered by material changes in the business: a large customer win or loss, a hiring freeze, or a macro shift significant enough that the prior forecast is no longer a useful planning tool.

**Rolling Forecast**
A continuously updated forward-looking financial projection that always covers a fixed horizon (e.g., the next 12 months). Updated monthly or quarterly as actuals come in and new information becomes available. Coexists with the fixed annual budget — it does not replace it. Better for operational decision-making; not a substitute for budget-based accountability.

**Run Rate**
An annualized projection based on a current period's performance, calculated by multiplying a period's result by the number of periods in a year (e.g., Q1 revenue × 4). Useful as a quick estimate but assumes the current period is representative of future performance — a dangerous assumption in businesses with seasonality, ramp effects, or significant growth trajectories.

**Scenario Planning**
The preparation of multiple complete financial plans under discrete, named alternative assumptions about the future. Each scenario is a coherent narrative with full P&L, balance sheet, and cash flow implications. Scenarios are not probability-weighted; they are used to bound the range of outcomes and test strategic decisions under different conditions. Common scenario names: downside, base, upside.

**Sensitivity Analysis**
A technique that varies one assumption (e.g., revenue growth rate) across a range of values while holding all other assumptions constant, to show how a single driver affects the financial outcome. Distinct from scenario planning: sensitivity varies one variable; scenarios vary many simultaneously within a coherent narrative. Sensitivity analysis answers "how much does our outcome change if this one input is wrong?"; scenario planning answers "what does the world look like under this set of conditions?"

**Three-Statement Model**
A financial model that integrates the income statement (P&L), balance sheet, and cash flow statement so that changes to any one statement flow through to the others correctly. The three statements are linked: net income flows to retained earnings on the balance sheet; changes in working capital and capex flow to the cash flow statement; cash on the cash flow statement reconciles to cash on the balance sheet. A valid FP&A model is three-statement by definition — working only from the income statement produces a forecast that cannot answer questions about cash, debt, or working capital.

**Top-Down Forecasting**
An approach where senior leadership sets the overall financial target first (e.g., 20% revenue growth), and FP&A or business unit managers backfill the operational assumptions required to achieve it. Faster and more aligned to strategic intent; can be disconnected from operational reality if targets are set without bottoms-up validation.

**Variance**
The difference between a planned (budget or forecast) value and an actual value. Expressed as a dollar amount, percentage, or both. Favorable variance (F): actuals better than plan (revenue above budget; expenses below budget). Unfavorable variance (U): actuals worse than plan. Variance analysis is the process of explaining the causes of material variances — decomposing the gap into volume, price, mix, and timing components and providing a forward-looking view.

**Variance Commentary**
The written explanation of why a variance occurred — not just its magnitude. Strong variance commentary isolates the causal driver (deal slippage, headcount timing, pricing change), segments the impact (which product, geography, or customer cohort), and provides a forward-looking view (expected timing of recovery or persistence of the gap). Weak commentary describes the variance in words without explaining it. The distinction between description and explanation is the most common failure mode in AI-generated FP&A content.

**Zero-Based Budgeting (ZBB)**
A budgeting approach in which every expense must be justified from scratch each cycle, rather than using the prior year's budget as a baseline and incrementally adjusting it. Forces explicit prioritization of every cost center; commonly used in cost-reduction programs or when entering a new strategic phase. Contrasted with incremental budgeting, which uses the prior period as a starting point. ZBB is more time-intensive and is typically applied selectively (high-cost functions, discretionary spend) rather than across the entire P&L.
<!--fold:432768@file path="sources.md" mode="644"-->
# FP&A Sources

Sources are organized by what they're most useful for. Prefer methodological frameworks over company-specific implementations.

---

## Methodology and frameworks

**Association for Financial Professionals (AFP) — FP&A Guide Series**
Practitioner-validated standards for FP&A methodology. The AFP FP&A Guide series covers forecasting practices, driver-based planning, and scenario planning in detail.
- FP&A resources: https://www.afponline.org/topics/FP-A
- FP&A Guide series (forecasting, driver-based modeling, scenario planning): https://www.afponline.org/publications-data-tools/reports/survey-research-economic-data/Details/fpa-guide
- Annual benchmarking data: https://www.afponline.org/publications-data-tools/reports/survey-research-economic-data

**Corporate Finance Institute (CFI)**
Covers the mechanics of FP&A in structured, searchable form — formulas, model structures, variance decomposition.
- Variance analysis: https://corporatefinanceinstitute.com/resources/accounting/variance-analysis/
- Rolling forecast methodology: https://corporatefinanceinstitute.com/resources/financial-modeling/rolling-forecast/
- Driver-based forecasting: https://corporatefinanceinstitute.com/resources/financial-modeling/driver-based-forecasting/
- Financial modeling best practices: https://corporatefinanceinstitute.com/resources/financial-modeling/financial-modeling-best-practices/

**Wall Street Prep — FP&A**
Strong on three-statement model mechanics and the link between P&L, balance sheet, and cash flow.
- FP&A overview: https://www.wallstreetprep.com/knowledge/fpa/
- Financial modeling fundamentals: https://www.wallstreetprep.com/knowledge/financial-modeling/

**CIMA / CGMA (Chartered Institute of Management Accountants)**
International standards body for management accounting. The CGMA framework covers planning, forecasting, and performance management methodology.
- Management accounting guidelines: https://www.cimaglobal.com/Professional-development/Research-and-development/
- CGMA tools and learning resources: https://www.cgma.org/resources/tools.html

---

## Planning technology and driver-based planning methodology

**FP&A Trends (Global FP&A Board)**
Research body producing annual surveys on FP&A practice — forecasting frequency, driver adoption, technology use, AI applications. More analytically rigorous than vendor content.
- Research and case studies: https://www.fpandatrends.com
- Annual FP&A survey: https://www.fpandatrends.com/research

**Workday Adaptive Planning — Methodology Guides**
Vendor documentation that is nonetheless useful for understanding driver-based planning structure and rolling forecast implementation patterns.
- Driver-based planning overview: https://www.workday.com/en-us/products/adaptive-planning/overview.html

---

## Practitioner perspectives

**The FP&A Guy**
Practitioner-authored. Useful for real-world variance analysis approaches, stakeholder communication, and forecasting technique discussion.
- https://www.thefpandaguy.com

**CFO Magazine**
Covers FP&A technology adoption, AI in forecasting, and practitioner case studies. More practitioner-facing than Gartner.
- https://www.cfo.com

**CFO Dive — FP&A**
News and analysis on FP&A practice, planning tool adoption, and AI applications.
- https://www.cfodive.com/topic/fpa/

---

## AI in FP&A

**Gartner — AI in Finance**
Research on forecast automation, anomaly detection, AI-generated commentary, and enterprise FP&A technology. Useful for situating AI applications within the broader enterprise finance landscape.
- Finance AI use cases: https://www.gartner.com/en/finance/topics/ai-in-finance

**McKinsey — AI in corporate finance**
Higher-level framing on AI-driven transformation of FP&A and financial planning functions. Less tactical than Gartner, more useful for strategic framing.
- https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai

---

## How to use these sources

**For structural FP&A methodology** (planning cycle, variance analysis framework, driver-based modeling): AFP FP&A Guide series is the practitioner-validated standard. Start here.

**For mechanics** (how variance decomposition works, how a three-statement model is structured, what driver-based forecasting looks like operationally): CFI and Wall Street Prep.

**For AI-in-FP&A framing** (what's being automated, where human judgment remains essential): FP&A Trends annual survey for current adoption data; Gartner for enterprise technology landscape; McKinsey for strategic framing.

**For current practitioner practice**: CFO Magazine and CFO Dive surface what finance teams are actually doing and what is failing in practice.

Do not cite vendor product pages as methodology sources. Workday, Anaplan, and Pigment publish methodology content that is useful, but it is written to support product adoption. The AFP and CIMA frameworks are methodology-first.
<!--fold:432768@end-->
PORTDOWN_EAECF7B5

# ── post ──
MARKER=$(awk '/^---$/ { f++; if (f==2) exit; next } f==1 && /^marker:[[:space:]]/ { sub(/^marker:[[:space:]]+/, ""); print; exit }' "$DEST")
[ -z "$MARKER" ] && { echo "seed: archive has no marker — corrupt" >&2; exit 1; }
awk -v m="$MARKER" -v outdir="$TARGET" '
  BEGIN {
    # Match <!--fold:<m>@file path="X"--> with an optional mode attr after
    # the path (fold emits  mode="644"  on executables).
    file_re = "^<!--fold:" m "@file path=\"([^\"]+)\"( mode=\"[0-9]+\")?-->$"
    end_re  = "^<!--fold:" m "@end-->$"
  }
  $0 ~ end_re { if (current) close(current); exit }
  $0 ~ file_re {
    if (current) close(current)
    line = $0
    sub(/^<!--fold:[^@]+@file path="/, "", line); sub(/".*$/, "", line)
    current = outdir "/" line
    dir = current; sub(/\/[^\/]*$/, "", dir)
    if (dir != current) system("mkdir -p \"" dir "\"")
    printf "" > current
    next
  }
  current { print >> current }
' "$DEST"
SEED_EXTRACTED=$(find "$TARGET" -type f -not -path "$DEST" 2>/dev/null | wc -l)
if [ "$SEED_EXTRACTED" = "0" ]; then
  echo "seed: archive contained no files — refusing to delete the source" >&2
  echo "  archive preserved at: $DEST" >&2
  exit 1
fi
rm -f "$DEST"

echo "" >&2
echo "✓ seed unpacked → $TARGET ($SEED_EXTRACTED files)" >&2
find "$TARGET" -type f | sort | while IFS= read -r _sf; do
  echo "  ${_sf#${TARGET}/}" >&2
done
echo "" >&2
if [ -f "$TARGET/SKILL.md" ]; then
  echo "This seed contains a skill (SKILL.md). Install it in your agent's skills directory." >&2
  echo "" >&2
fi
echo "Install the seed skill if not already installed:" >&2
echo "  https://seed.show/skill" >&2
echo "" >&2
echo "Publisher prompt:" >&2
sed 's/^/  /' >&2 <<'__SEED_PROMPT_END_AC1F2B__'
Use this as grounding context for FP&A work: read README.md for the planning cycle, variance framework, agent failure modes, and what AI is/isn't changing; consult glossary.md for precise term definitions; use sources.md to find methodological references.
__SEED_PROMPT_END_AC1F2B__
exit 0