Skip to content

JPeetz/agent-skills / skills/finance-ops/saas-metrics-coach

saas-metrics-coach

Verified

Use this skill when analyzing SaaS business metrics, building financial models, or preparing investor metrics. Activates on "SaaS metrics", "ARR", "MRR", "churn", "LTV", "CAC", "unit economics", "recurring revenue", "financial model", "runway", "burn rate", "net revenue retention", "NRR", "quick ratio", or "investor metrics". Covers revenue metrics (ARR/MRR/NRR), churn and retention, LTV/CAC ratio, unit economics, cash flow and runway, cohort analysis, benchmark comparisons, and board/investor metric packs.

FileKindSize
CHANGELOG.mdmarkdown1.5 KB
LICENSEother1.0 KB
SKILL.mdskill md21.5 KB
evals/evals.jsonconfig8.3 KB
references/sources.mdmarkdown5.5 KB
scripts/metrics_calculator.pyscript28.8 KB

evals/evals.json

View on GitHub
[
  {
    "id": "pos-01",
    "type": "positive",
    "name": "Full metrics pack from raw subscription data",
    "description": "Test that a raw data dump produces the complete computed set with definitions and a bridge check, not just a list of numbers.",
    "test_input": "Here is our H1 snapshot: MRR went from 100,000 to 128,000 over six months (Jan-Jun 2026). New business 30,000, expansion 9,000, reactivation 2,000, contraction 4,000, churn 9,000. Customers 200 to 214, 30 new, 16 churned. Revenue 660,000, COGS 132,000, sales & marketing 180,000, variable costs 33,000. Cash 4.2M, burning 100,000/month. Mid-market segment. Run scripts/metrics_calculator.py and give me the whole pack.",
    "expected_output": "ARR 1.2M to 1.536M, NRR 96%, GRR 87%, quick ratio 3.00, gross revenue churn 13% for the period and ~2.3%/month, blended CAC 6,000, LTV/CAC 2.91, CAC payback 15.0 months, burn multiple 1.79, runway 42 months — each with the definition used, a benchmark band and the flags the calculator raised.",
    "assertions": [
      "ARR is reported as MRR x 12 (1,200,000 to 1,536,000), not as revenue x 2 or an annualised single month",
      "The MRR bridge is stated and shown to close (begin + new + expansion + reactivation - contraction - churn = end)",
      "NRR is 96% and the response says it excludes new business and reactivation",
      "Monthly churn is derived by de-annualising the period rate, not by dividing 13% by 6",
      "LTV is computed on gross-margin dollars (80%), with the lifetime assumption stated",
      "CAC payback is 15.0 months and compared against the mid-market norm rather than a single universal number",
      "Both the blended and paid CAC variants are reported or the paid variant is explicitly marked unavailable",
      "Flags are surfaced with severity (NRR below 100%, LTV/CAC below 3.0)",
      "No metric is reported without its definition or time basis"
    ]
  },
  {
    "id": "pos-02",
    "type": "positive",
    "name": "NRR dispute — definition and diagnosis",
    "description": "Test that an ambiguous NRR claim is resolved by definition first, then diagnosed with cohorts, and that the response refuses to pick a definition silently.",
    "test_input": "My board deck says NRR is 112% but my CFO says it's 94%. We're a $4M ARR SMB product with lots of reactivated accounts and a 22% annual logo churn. Who is right and what do I tell the board?",
    "expected_output": "An explicit statement that both can be arithmetically correct under different definitions: the 112% figure almost certainly includes new logos or reactivation, the 94% excludes both. The canonical NRR formula is restated, both variants are shown, and the recommendation is the stricter variant plus a note on which one is being used. Cohort and logo-churn evidence is requested before the trend is called improving.",
    "assertions": [
      "States that NRR excludes new business and says which variant the two figures likely represent",
      "Gives the exact formula used: (begin + expansion - contraction - churn) / begin",
      "Reports both NRR and NRR including reactivation rather than choosing silently",
      "Notes that 22% annual logo churn is inconsistent with a genuinely healthy 112% NRR in an SMB base",
      "Recommends labelling the definition on the board slide and restating history if the definition changes",
      "Asks for cohort-level retention before declaring the trend improving"
    ]
  },
  {
    "id": "pos-03",
    "type": "positive",
    "name": "CAC audit — headcount excluded",
    "description": "Test that the coach catches a flattered CAC built without S&M headcount and recomputes payback, rather than accepting the reported ratio.",
    "test_input": "We report CAC of $2,400 and LTV/CAC of 9x. Last year: $480k on ads, $1.2M on S&M payroll, 200 new customers. ARPA is $450/month, gross margin 78%, monthly churn 1.5%. Is 9x good?",
    "expected_output": "A recomputation showing paid-only CAC of $2,400 but fully-loaded CAC of $8,400 once $1.2M of S&M payroll is added, so on gross-margin LTV (~$23.4k) LTV/CAC is about 2.8x and payback is about 24 months — a retention or pricing problem, not under-investment. The response declines to celebrate the 9x.",
    "assertions": [
      "Identifies that the reported CAC excludes S&M headcount",
      "Recomputes fully-loaded CAC as ($480k + $1.2M) / 200 = $8,400",
      "Recomputes LTV on gross-margin dollars (about $23.4k, lifetime 1/1.5% = 66.7 months)",
      "Recomputes LTV/CAC (~2.8x) and CAC payback (~24 months) and shows they fall below the 3.0 and 12-month references",
      "Does not call a 9x ratio under-investment before the inputs are validated",
      "Recommends an explicit CAC definition (blended, paid, fully-loaded) for the board pack",
      "Provides a sensitivity statement on what the ratio becomes if churn rises"
    ]
  },
  {
    "id": "pos-04",
    "type": "positive",
    "name": "Runway and burn multiple for an investor update",
    "description": "Test that runway is computed from net burn with committed outflows and planned hiring, and that burn multiple is computed over a period rather than a month.",
    "test_input": "Investor update due Friday. We have $3.1M in the bank, burned $310k net last month, $280k and $340k the two months before. Net new ARR last quarter was $420k, S&M spend $390k. Growth was 55% YoY and we're planning to hire 3 engineers in Q1. What do I write for runway and efficiency?",
    "expected_output": "Net burn of ~$310k/month on a trailing 3-month average, runway of about 10 months on current burn and less once the hiring plan is modelled, a burn multiple of ~2.2 for the quarter, and the recommendation to state the hiring-adjusted runway plus a fundraising trigger now.",
    "assertions": [
      "Uses a trailing three-month average net burn (~$310k) rather than one month",
      "Computes runway as cash / net burn and states the result in months",
      "Adjusts or explicitly annotates runway for the planned engineering hires",
      "Computes burn multiple as quarterly net burn / quarterly net new ARR, not monthly",
      "Flags that sub-12-month runway is a fundraising trigger",
      "Reports magic number or notes that S&M spend alone does not equal CAC",
      "Warns that a flat-burn runway is a lower bound for a hiring company"
    ]
  },
  {
    "id": "neg-01",
    "type": "negative",
    "name": "Revenue recognition and month-end close",
    "description": "Near-miss: it mentions recurring revenue and ARR, but the request is an accounting-policy question, not metrics analysis.",
    "test_input": "We signed a 3-year contract with annual prepayment. Should I recognise the full amount as ARR at signing, and how do I book the deferred revenue in our month-end close so the auditor is happy?",
    "expected_output": "Should NOT load this skill's metrics workflow. The correct response routes to revenue-recognition/accounting guidance (ASC 606 / IFRS 15), explains that ARR is a management metric distinct from recognised revenue and bookings, and does not produce a metrics pack.",
    "assertions": [
      "Does not produce a SaaS metrics pack or run metrics_calculator.py",
      "Routes to accounting / revenue-recognition guidance",
      "May clarify that bookings, billings, recognised revenue and ARR are four different lines",
      "Does not attempt to set revenue-recognition policy"
    ]
  },
  {
    "id": "neg-02",
    "type": "negative",
    "name": "Pricing and packaging decision",
    "description": "Near-miss: the prompt contains 'ARR' and 'revenue', but the decision is which pricing model to adopt — pricing-strategist territory.",
    "test_input": "We're at $2M ARR on flat-rate plans and want to move to usage-based pricing with good/better/best tiers. What should we charge and how should we package it?",
    "expected_output": "Should NOT run the metrics workflow. The correct response routes to the pricing skill: model selection, willingness-to-pay range and tier packaging. Metrics may be referenced as inputs (ARPA, expansion potential) but the deliverable is a pricing model and packaging, not a financial metrics pack.",
    "assertions": [
      "Does not emit a metrics pack, benchmark tables or run metrics_calculator.py",
      "Routes to pricing/packaging guidance",
      "Does not propose a single price without willingness-to-pay evidence",
      "Any metric cited is used as an input to pricing, not reported as the deliverable"
    ]
  }
]
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexUnverifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall