Sample data#
Sample data gives you coherent lakes without using customer information. Each persona is a fictional company with deterministic records, cross-source identities, lifecycle history, and dashboard-ready results.
Personas#
| Persona | Story | Representative sources |
|---|---|---|
| Skycrane | Sales-led B2B software | HubSpot, Stripe, Intercom, PostgreSQL |
| Arecibo | Usage-billed API business | Product usage, billing, support |
| Flightsuit | Direct-to-consumer commerce | Shopify, Klaviyo, Facebook Ads, Google Ads |
| Moonshot Scholars | Education nonprofit | Salesforce, Google Sheets, Airtable |
| Static Fire | Consumer iOS product | Apple App Store, MySQL, Amplitude |
| Heatshield | Enterprise software | Pardot, Pendo, Zendesk, Jira, NetSuite, Gainsight, Canny, Microsoft Ads, and others |
| Collapsar Games | Mobile game studio | 8 product, commerce, and marketing sources |
Open /config/sample to explore the published profiles and their dashboards. The public demos use checked-in results while browser-side lake execution is being rolled out.
Determinism#
Every value is a pure function of persona, entity, stream, and date. Running a generator twice for the same date produces value-identical data. A later daily run appends only its new operations.
The full persona lakes use the same writer and catalog contracts as product data. Schema manifests derive from source definitions where a connector is fixed-schema.
Safety#
Published domains end in .example, which cannot be delegated. Phone numbers use 555-0100 through 555-0199. IP addresses come from RFC 2544 and RFC 5737 ranges. Names are screened, but structural address safety does not depend on a company-name blocklist.
No sample is derived from customer data. Published datasets and their license text are MIT licensed.
Use the examples#
Documentation defaults to Skycrane for B2B examples and Flightsuit for commerce. Reusing the same tables across pages lets queries compound instead of introducing a new fictional world for every feature.