Free test data generator

The standard commerce star schema — customers, products, orders, order lines, payments and returns — with order totals that reconcile against their lines.

• customers — One row per person who has an account.
• products — The catalogue. Cost is included so margin is computable.
• orders — One row per order. Totals reconcile against order_items.
• order_items — Order lines. line_total = quantity × unit_price, less line discount.
• payments — What was actually taken. A captured payment equals the order total.
• returns — Returned lines. Never more than was bought, never before the order.

Recurring revenue with trials, upgrades and churn — accounts, seats, subscriptions, plan changes, invoices, usage telemetry and support tickets.

• accounts — One row per paying organisation.
• users — Seats. Belongs to an account, never created before it.
• subscriptions — The contract. ended_at is null while it is still running.
• plan_changes — Upgrades and downgrades, with the MRR movement each caused.
• invoices — One per billing period. None is dated after its subscription ended.
• usage_events — Product telemetry. Only occurs while the subscription is live.
• tickets — Support. Resolution time follows priority; satisfaction follows resolution time.

Patients, providers, encounters, diagnoses, medications and observations, with age-appropriate conditions and plausible vitals — and codes from an invented scheme, not ICD or SNOMED.

• patients — People. No real person is described by any row.
• providers — Clinicians. Specialty drives which encounters they take.
• encounters — A visit. discharged_at is always at or after admitted_at.
• diagnoses — Coded against the pack's own SDS- scheme, not ICD or SNOMED.
• medications — Prescriptions, tied to the condition that prompted them.
• observations — Vitals and labs, with an abnormal flag derived from the value.

An append-only ledger that balances: customers, accounts, cards, merchants and transactions whose running balance reconciles exactly, with salary cycles, paired transfers and labelled fraud.

• customers — Account holders.
• accounts — current_balance equals the balance_after of the last transaction.
• cards — Card numbers are masked; nothing here passes a Luhn check, deliberately.
• merchants — Where the money went.
• transactions — The ledger. balance_after is computed forward and always reconciles.

An event stream with geography — hubs, vehicles, shipments and a scan trail per parcel, with clustered delays, failed delivery attempts and a status always derived from the last scan.

• hubs — Sorting facilities. Capacity drives how often one falls behind.
• vehicles — The fleet, each based at a hub.
• shipments — One parcel. status always matches the last scan on the trail.
• scans — The trail. Strictly increasing in time within a shipment.
• exceptions — Why a shipment went wrong, tied to the scan that recorded it.

Slowly-changing records: departments, employees with a real management tree, contiguous compensation history, performance reviews and leave.

• departments — Cost centres.
• employees — People. manager_id points at someone hired earlier who had not left.
• compensation — Contiguous, non-overlapping intervals. Exactly one current row per employee.
• reviews — Performance cycles. A poor final review makes an exit more likely.
• absences — Leave taken. Always within employment, never on a weekend.