The company you are building,in data, before you have it.
You have a product and a schema, not three years of customers. Tell us the business and the numbers it should show. We build every table behind it, connected, with those numbers held exactly, and deliver the files with a certificate of what was checked.
- Your schema, followed exactly
- Your numbers, held exactly
- No real customer data
- New
- Expansion
- Contraction
- Churn
A revenue dashboard on a demo dataset built from one sentence: 11 tables, 14,341 rows. Every figure is read from its tables.
What you send, what comes back.
- What it is
- Custom synthetic datasets, built to order by Misata Studio
- Who it is for
- Startups with a product and a schema, but no customer history yet
- You send
- The business, your schema if you have one, and the numbers the data should show
- You get
- Connected tables in your names, each stated number held, a certificate of what was checked, and the recipe
- Schemas read
- SQL DDL, Prisma, Django, SQLAlchemy, Rails schema.rb and dbt projects
- Delivered as
- SQL for Postgres, MySQL, SQL Server, Oracle, BigQuery, Snowflake or ClickHouse; Parquet, CSV, SQLite, DuckDB or dbt seeds
- Size
- Up to about 10 million rows
- Real data needed
- None, at any point
- First reply
- Aimed within one business day: questions, a sample schema, a price and a delivery date
Five places a startup needs data it does not have yet.
The product demo
An account with a year of activity behind it: dashboards with something to show, a churn spike your feature catches, a plan mix that makes the pricing page make sense. Built to the walkthrough you want to give.
Trial and sandbox accounts
New users land on a workspace that already shows the product working, in their own industry, and every account tells the same story because the same recipe builds it.
A development and staging database
Built from your Prisma, Django or SQLAlchemy models, Rails schema, dbt project or SQL DDL, with every key matching. No copy of production on anyone's laptop.
Testing at the size you are heading for
The volumes you expect next year, up to about 10 million rows, so slow queries and heavy pages show up before your customers find them.
A company for your agent to work in
CRM, billing, support and the books sharing the same customers and orders, so an AI agent can be built and shown without anyone's real records.
Say the story. Get every table behind it.
The numbers are agreed before a row is built, and measured on the rows after.
- 01 · You send
A seed-stage B2B SaaS for dental clinics with 18 months of history. MRR grows from $15,000 to $60,000, churn runs at 3% a month, and the Pro plan brings 40% of revenue.
- 02 · We confirm
- History
- 18 months
- MRR, first month
- $15,000
- MRR, last month
- $60,000
- Monthly churn
- 3%
- Pro plan, share of revenue
- 40%
- 03 · You get
- Your tables, in your names, in the format you use
- Each number you stated, held in every month it covers
- A certificate of what was checked, shareable by link
- The recipe, so the same data can be built again
An example request. Files come as SQL for Postgres, MySQL, SQL Server, Oracle, BigQuery, Snowflake or ClickHouse, or as Parquet, CSV, SQLite, DuckDB or dbt seeds.
Four ways to fill a database before you have customers.
| Fits your schema | Every table joins | Shows the numbers you choose | No real customer data | Keeps up as you change | |
|---|---|---|---|---|---|
| A public sample dataset | Someone else's tables | Within its own tables | Whatever it happens to hold | Often real people; check the licence | Mapped again by hand |
| Faker or a seed script | You write it | Where someone wired them | Random, not a story | Invented | Breaks on migrations |
| Rows asked of a chatbot | Close, for a table or two | Break past a few tables | Plausible, not held | Invented | Different every time |
| A custom Misata dataset | Built from your schema | Every key checked | Held exactly, on a certificate | Built from scratch | Rebuilt from its recipe |
Selling to customers you already have? Filling a sales demo or a trial account is covered on demo data.
Checked on the finished tables, not promised.
The same guarantees as every Studio dataset. How we verify.
Every table joins
Every foreign key points at a row that exists, and rows keep their order in time: no invoice before its subscription.
Your numbers, held exactly
Each number you state is measured on the delivered rows, month by month, and listed on the certificate as met or not.
Rebuilt when your schema moves
Every dataset is built from a recipe. A new column or table is a rebuild with the same story, not a seed script to patch.
No real data, at any point
Built from your description and your schema. Never send customer records or production dumps.
Tell us what the data has to show.
We aim to reply within one business day with questions, a sample schema, a price and a delivery date.
- 01The business
What the product does and who uses it, in a sentence or two.
- 02Your schema, if you have one
Paste the DDL or models, or say you have none and we propose one.
- 03The numbers it should show
Growth, churn, plan mix, seasons: whatever the demo or test depends on.
- 04Size and format
Rows or months of history, and the database or files you load it into.
Rather do it yourself? Describe it in Studio.
What startups ask first.
How can a startup get demo data before it has customers?
How is this different from Faker or a seed script?
We do not have a schema yet. Can you still help?
Can you build from our Prisma, Django, Rails or dbt models?
Which databases and file formats do you deliver?
How large can the dataset be?
Can we use it in a demo for prospects or investors?
Our schema changes every week. What happens then?
Do we need to send real data?
How much does it cost, and how fast is it?
What can you not do?
Updated 2026-10-29. See also known limitations and how your data is handled.
Show what your product does, before your customers do.
Your schema, the story you want to tell, every table connected and every number you state held exactly.