THE CUSTOMER CUBE ARTIFACT

Connect your AI client to Chargebee, ask one question in plain language, and it reads your full subscription history, converts every currency to USD, and builds a customer-level ARR cube spanning nine quarters. Every paying account, sorted by product line. No exports, no spreadsheets, no waiting on a data team.

The manual alternative is a quarterly grind: pull the subscriptions, reconcile the currencies, rebuild the pivot, recompute growth and retention by hand. It hands you a headline number, not a view. It rarely surfaces the vertical whose net retention is quietly carrying the whole book (Financial Services here, near 136%), the account that has gone flat for nine straight quarters on a 30% discount, or the new logo that ramped from zero to seven figures inside two years.

The output is one explorable artifact with three views.

Summary lays out total ARR, growth, and gross and net retention, each expandable by segment, region and industry, alongside discount bands and ranked tables for top accounts, expansion, new sales and churn.

Insights reads the cube back as plain-language strengths, risks, signals to watch and open strategic questions, each with a suggested next step.

Raw Data is the full, sortable, filterable customer table. All produced in a single conversation.

Anonymized sample shown. Company names, domains and figures here are fictional and for illustration only.

02Why this works

The history is
already there.

Most platforms record only the current state of a subscription. The commercial history (when it started, when it ended, when terms last changed, what it was worth at each point) lives elsewhere, usually assembled by hand on request.

Chargebee is built differently. The full event chain is stored on every subscription record, which is what makes nine-quarter time-series analysis possible without a data warehouse or a data team. Connect an AI client through the Model Context Protocol and that history becomes queryable in plain English. One question retrieves the records, converts currencies, sorts products, and formats the output. No SQL. No scheduled export. No analyst in the loop.

03 The exact prompt

Paste it into your AI client to build your own ARR cube.

Show the full prompt+

Objective: Using the Chargebee MCP connector, build a customer-level ARR cube showing every paying customer's recurring revenue by quarter from Q1 2024 through Q1 2026. This will be used as a board-level revenue intelligence report.

Step 1 — Pull subscriptions from Chargebee
Make paginated calls to list_subscriptions across three passes to capture all relevant revenue:
  • Pass 1: status = active
  • Pass 2: status = non_renewing
  • Pass 3: status = cancelled, filtered to cancelled_at on or after January 1, 2024

For each pass, use limit = 100 and paginate using next_offset until all records are retrieved. Only process subscriptions where mrr > 0 (exclude free, trial, and zero-revenue accounts).

Step 2 — Extract the following fields from each subscription
  • subscription.customer_id
  • subscription.mrr (in cents for most currencies; do not divide for zero-decimal currencies: JPY, KRW, IDR, etc.)
  • subscription.currency_code
  • subscription.status
  • subscription.started_at
  • subscription.cancelled_at
  • subscription.current_term_end
  • subscription.subscription_items[].item_price_id
  • subscription.subscription_items[].amount
  • subscription.discounts[]
  • customer.billing_address.company
  • customer.email (extract domain from everything after @)

Step 3 — Aggregate by customer
Group all subscriptions by customer_id. For each customer:
  • Sum mrr across all active subscriptions
  • Convert to ARR in USD: ARR (USD) = (mrr ÷ 100) × 12 × FX rate
  • FX rates (May 2026): USD 1.000 · EUR 1.082 · GBP 1.273 · AUD 0.647 · CAD 0.729 · INR 0.012 · SGD 0.741 · JPY 0.00671 · MXN 0.052 · BRL 0.193 · AED 0.272 · CHF 1.131. Use 1.0 for any unlisted currency.
  • Use the currency and FX rate of the customer's highest-MRR subscription as representative.

Step 4 — Calculate quarterly ARR
For each of the 9 quarters, a subscription is active at quarter-end if:
  started_at ≤ quarter_end AND (status = active OR cancelled_at > quarter_end OR current_term_end > quarter_end)

Quarter-end Unix timestamps:
  Q1 2024: 1711929599 · Q2 2024: 1719791999 · Q3 2024: 1727740799 · Q4 2024: 1735689599
  Q1 2025: 1743465599 · Q2 2025: 1751327999 · Q3 2025: 1759276799 · Q4 2025: 1767225599
  Q1 2026: 1775001599

Step 5 — Classify each customer's primary product
Using item_price_id from the highest-MRR subscription (first match wins):
  • Contains impl or implementation → Exclude (one-time charge)
  • Contains revrec → RevRec
  • Contains retention → Retention
  • Contains receivabl → Receivables
  • Contains cpq → CPQ
  • Contains enterprise, performance, scale, rise, finance suite, rgm, usage-based, ubb, or starter → Billing
  • Everything else → Other ARR

⚑ Customize this step — the product names above (RevRec, Retention, Receivables, CPQ, Billing) are specific to this Chargebee instance. Replace the keyword matches and category labels with your own product line names before running.

Step 6 — Calculate total discount
Blended discount rate = total discount amount ÷ MRR before discounts (from subscription.discounts[] on the primary subscription). Express as a decimal.

Step 7 — Output one row per customer, sorted by Q1 2026 ARR (USD) descending
Exclude any customer where Q1 2026 ARR = $0.

Columns: ID · Domain · Account Name · URL · Product · Current Plan Handle · Currency · FX Rate · Total Discount · Q1 2024 Plan ARR through Q1 2026 Plan ARR (all in USD).

Format ARR as whole dollars with commas (e.g. $1,390,939). Display $0 as a dash (—). Export as a CSV file.

Customize Step 5 before you run this. The product names above (RevRec, Retention, Receivables, CPQ, Billing) reflect our company's product line. Replace the keyword matches and category labels with your own before running the prompt.

Build your Board-ready Customer Cube artifact with Chargebee MCP

The history isalready there.

The history is
already there.