Scratch Data
Scratch Data, formerly known as Hotswap and Tasker, is a B2B company specializing in finance and accounting solutions, offering scalable data analytics plans and an open-source platform with comprehensive API access.
Company History
Scratch Data, formerly known as Hotswap and Tasker, is a company in the B2B industry, specifically within Finance and Accounting. The company is part of the Y-Combinator S21 batch. It operates out of New York, NY, and supports remote operations, serving regions including the United States of America and Canada. Scratch Data is described as an 'Open-Source Snowflake,' focusing on open-source solutions for data processing.
Services
Scratch Data offers various service plans targeted at different stages of company growth. The self-hosted open-source plan with API access and all database connectors is available at $0 per month. The Startup plan, priced at $24 per month, includes a hosted analytics pipeline, dashboard, data sharing with one-time links, 100 GB of data transfer, and email support. The Growth plan is designed for growth-stage companies, offering 1 TB of data transfer, bulk data pricing, and a private Slack channel at $149 per month. They guarantee to unlock data in 30 minutes with a first-month-free offer if not connected within a single Zoom call.
Technology and Platform
Scratch Data’s platform is built on top of Clickhouse and automates all aspects of managing an analytical database, including server configuration, data ingestion, queries, replication, and sharding. The platform provides a RESTful API for data streaming to databases, automatic table setup, schema migrations, and supports webhooks for automatic data sending from services like Stripe and Shopify to the warehouse. It allows the creation of realtime analytics products by querying data via an API without needing database drivers, credentials, or connection management.
Product Features
Scratch Data’s platform features include the ability to handle large-scale data, having processed 15 terabytes and ingested 5 billion rows since launch. The average integration time with the platform is 13.2 minutes. The API responses are returned in milliseconds, providing a unique ID for each ingested data piece. It supports creating one-time links to datasets in formats like JSON, CSV, Excel, Parquet, and enables the creation of API endpoints that restrict which rows can be queried, facilitating a user-friendly developer experience.