Misata Logo
Misata Studio
Custom datasets

A custom dataset, built to your schema and your numbers.

Tell us the dataset you need: the tables, the volumes, the rates and the rules that must hold. We reply with a proposal and a sample schema, build it with Misata Studio’s engine, and deliver the files with a certificate of what was checked. No real data is needed at any point.

Every foreign key checked Your numbers applied exactly No real data needed

Need something today? Start from a free dataset.

Illustration of a data workshop stamping a seal of approval on finished datasets
Built with Studio’s engineDelivered with a certificate

What people ask us for

A demo database for a prospect

Their industry, their region, their numbers: a revenue curve, a plan mix, a churn rate that tells the story the demo needs.

Labelled fraud or AML data

Legitimate behaviour by segment with laundering or fraud patterns planted at an exact rate, so a detector's precision and recall can be measured.

A development database from your DDL

Your tables and names followed exactly, filled with believable values, every key matching, up to about 10 million rows.

An evaluation set with an answer key

Data with known answers planted in it, and deliberate errors listed row by row, for testing a model, an agent or a cleaning pipeline.

Data for something that does not exist yet

A new market, a future year, a product before launch: written for that world, with real facts researched and linked where they exist.

What you get

The same guarantees as every Studio dataset, checked on the finished tables. How we verify.

Every relationship checked

Every foreign key points at a row that exists, checked on the finished data, across every table.

A certificate of what was checked

Each number and rule you asked for, whether it was met, and how it was measured. Shareable by link.

No real data involved

Built from your description or schema. Never send us customer records or production dumps.

Files for your stack

SQL for Postgres, MySQL, SQL Server, Oracle, BigQuery or Snowflake, Parquet, CSV, SQLite, DuckDB, dbt seeds and more.

How it works

  1. 01.
    Describe what you need: Your tables or DDL, the volumes, and the numbers and rules that must hold. As much or as little as you have.
  2. 02.
    Get a proposal: We reply by email with questions if we have them, a sample schema, the price and a delivery date.
  3. 03.
    Receive the data and its certificate: Built with Studio's engine, checked, and delivered in the formats you asked for. Changes are made by regenerating from the recipe.
Request

Describe the dataset you need

Give as much detail as you have. We aim to reply within one business day with questions, a sample schema and a proposal.

Describe tables, relationships, row counts, distributions, or constraints you care about.

We will respond within 24 hours with a schema preview and a quote. Bespoke datasets start at $50. Your email is never shared or added to a list.

Questions

Do I need to send you real data?
No, and please do not. We build from your description, your schema (SQL DDL, DBML, Prisma or a CSV header) and the numbers you give. No production rows are needed at any point.
How is this different from using Misata Studio myself?
It is the same engine and the same checks. Studio is self-serve and quick for most requests. Ask us when the schema is large, the rules are specialised, you need several related datasets, or you would simply rather hand it off.
How much does a custom dataset cost?
Each request is quoted in the proposal, before any work starts. If Studio can already do what you need on its own, we will say so, and you can do it yourself on Studio's credits.
How fast will I hear back?
We aim to reply within one business day. The delivery date depends on the request and is agreed in the proposal.
What can you not do?
We do not mask or copy production data, and we do not learn from your real records. Data is invented to match what you describe, so it is not a forecast or evidence about real people. The known limitations page lists the rest.

Updated 2026-09-24. See also known limitations and how your data is handled.

Request a dataset