SaaS Metrics from Stripe, Paddle & Lemon Squeezy

    1

    MRR, churn, NRR and cohort retention from your own billing export, calculated by a script, not guessed.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    SaaS Metrics from Stripe, Paddle & Lemon Squeezy

    Example session with this skill installed

    Here's our Stripe export. What were MRR, churn and net revenue retention over the last six months? It's for next week's board update.

    • Read your context and instructions
    • Compiled the saas metrics from

    Plan: check the export with --inspect, then report May to October 2026 in USD: MRR, movements, churn, retention and cohorts.

    1. Checking the export

    The invoices export failed the check

    "period_ends_on_or_before_date_pct": "100.0"
    

    Every period ends on or before its invoice date, so the periods look back one cycle [Look-back invoice periods]. Run as it is, it reports October MRR of 0.00, with all 1102.20 counted as churned. I asked for the invoice line items export instead. It passed: 0.0 look-back, 63 MRR lines, amount - discount, and 1 row skipped (an open invoice, as expected).

    2. October 2026 headline (USD)

    MRRARRCustomersARPUNet new MRR
    1201.2014414.409133.47-99.00

    3. What moved MRR [MRR movements]

    MonthNewExpansionContractionChurnedReactivationMRR
    2026-0799.000.000.0099.000.00742.00
    2026-08159.20150.000.0049.000.001002.20
    2026-0999.00200.00100.000.0099.001300.20
    2026-100.000.000.0099.000.001201.20

    July: the new annual plan counts as 99.00 a month [Normalizing billing intervals]. Its 500.00 setup fee is one-off revenue, outside MRR [One-off charges].

    • September: the reactivation is the customer who paused in July.

    October: the churn is ****0008, whose invoice was refunded in full (99.00 in refunds and credits) [Refunds and credits].

    4. Churn and retention

    • Logo churn [Logo churn]: 12.5% in July and in August, 0.0% in September, 10.0% in October.
    • Revenue churn [Revenue churn]: 7.6% in October.
    • May to October [Net revenue retention (NRR)]: the customers paying 693.00 at the end of May (7 of them) pay 114.7% of it now, so NRR is 114.7%. GRR is 64.2% [Gross revenue retention (GRR)]. Upgrades from existing customers outweighed what was lost.

    Cohorts: May to September have one customer each. Four are at 100.0%, and May is at 0.0% in October. These are single accounts, not rates.

    Next steps

    • For a 12-month NRR, export line items back to October 2025 and rerun with --from 2025-10 --to 2026-10.
    • Leave November out until it has closed. A partial month shows renewals that haven't happened yet as churn.
    • These are management metrics, not recognized revenue or advice. Check revenue recognition and anything about valuation with a qualified professional.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Drop in a Stripe, Paddle or Lemon Squeezy export, or any CSV, and get board-ready SaaS metrics. A bundled standard-library Python script computes every number, so nothing leaves your machine. You get monthly MRR and ARR with annual plans normalized and discounts applied; MRR movements split into new, expansion, contraction, churned and reactivated; customers, ARPU, logo and revenue churn, gross and net revenue retention, and a cohort retention table by first paid month. It handles the traps that skew dashboards: trials, refunds, proration, paused and cancel-at-period-end subscriptions, one-off charges, mixed currencies and months with no starting revenue. Customer ids and emails are masked, and every figure cites a written definition.

    Requires Python 3. No currency conversion or tax recognition. MRR is a month-end snapshot. Snapshots exports cannot show historical price changes. Proration is day-normalized.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 9 days ago

    • Free with an Agensi account

    Requires

    Python 3

    Permissions

    Terminal / Shell
    Read Files
    Write Files

    Runs the bundled scripts locally on the files the user names. Writes output only inside the user's project, after asking, and makes no network requests.

    Listed9 days ago

    What's inside

    Frequently Asked Questions

    This skill uses a local Python script to process your billing CSVs directly on your machine. Unlike web-based dashboards, your sensitive customer and revenue data never leaves your local environment or gets uploaded to a third-party server.

    Run the --inspect step first: it shows which columns were detected and any problems with the export. If a column was missed or named differently, map it with --map field=Header (for example --map amount=Net). That lets it read Stripe, Paddle, Lemon Squeezy or any billing CSV with a customer, a date, an amount and a period or billing interval.

    No, this skill is designed for ease of use and uses only the Python standard library. You simply need Python 3 installed on your system; there are no complex external libraries or dependencies to manage.

    LLMs are unreliable at long arithmetic and proration. Here a deterministic Python script computes every number, and the agent only explains the script's output and cites the definitions in reference.md. The script ships with a self-test of hand-checked cases (annual plans, trials, upgrades, refunds, churn and reactivation).

    The skill is three parts: SKILL.md (the agent's instructions), scripts/saas_metrics.py (the calculator, Python standard library only) and reference.md (each metric's formula and the export traps it handles).

    Currencies are never mixed: every figure is reported separately for each currency, and there is no conversion. Tax is taken off when the export has a tax column, and a total column is treated as already net of discounts. Pass --currency for exports with no currency column.

    New in Finance & Accounting