From a description to a loaded database, in five steps.
Hand-written seed data usually means broken foreign keys, totals that do not add up, and hours spent debugging staging. This guide shows how to get connected, realistic data out of Misata Studio, with every number you state applied exactly and checked.
How to get a dataset you can trust
Click a step to see what to do, an example, and what to check.
Describe it with numbers
Say what the business is, and state every figure that matters
A sentence is enough to start, but every number you write becomes a requirement Studio applies exactly and checks: counts, shares, rates, a date window, a revenue curve, the country. What you leave out is written for the business you describe, and listed as an assumption so you can see it.
- Write shares as numbers (60/30/10), not 'mostly free'. Numbers are applied; adjectives are interpreted.
- Name the country or city: names, phones, postcodes, currency, holidays and time zone follow it.
- Give a date window or an era, so every timestamp lands inside it.
A B2B SaaS company in Germany with 2,500 accounts
on three plans: Free 60%, Pro 30%, Enterprise 10%.
Each account has 3 to 25 users.
MRR grows from €40,000 in January 2025 to €150,000
in December 2026, with 4.2% monthly churn.Start from one of these
Accounts, cards and an exact fraud rate
Customers, checking, savings and credit accounts, and a card transaction ledger. Balances follow from the transactions, and the fraud rate is exact and lands on the riskier transactions.
“A neobank in the UAE with 10,000 customers, checking, savings and credit accounts, 150,000 card transactions over one year across five merchant categories, and 1.8% of transactions flagged as fraud.”
Accounts, seats, invoices and churn
Accounts on plans, users with roles, subscriptions, MRR movements and invoices. The churn rate is exact, and ARR and current MRR always agree with the movements.
“A B2B SaaS platform with 2,000 accounts on Free, Pro and Enterprise plans, users with admin, member and viewer roles, monthly invoices, and 4.2% monthly churn over 12 months.”
Patients, encounters and ICD-10 diagnoses
Patients, physicians, encounters, diagnoses with real ICD-10-CM codes that agree with each patient's sex, and claims with the denial rate you set. No real patient is involved.
“An outpatient network with 5,000 patients, 40 physicians across 8 specialties, 25,000 encounters with ICD-10 diagnoses, and a 6% claim denial rate.”
Sensor readings that drift before a failure
Machines that wear out, sensors that drift by failure mode, the gaps and spikes real telemetry has, and failures labelled with the true time remaining.
“A cold-chain fleet of 300 refrigerated trucks with temperature and compressor readings every 15 minutes for three months, and compressor and door-seal failures.”
Rather have us build it?
If your schema is large, your rules are specialised, or you simply want someone to do it, describe what you need. We reply with a proposal and a sample schema, build it with Studio’s engine, and deliver the data with its certificate.
- Every foreign key checked
- A recipe that rebuilds it
- Files for your database, with a load command
- A certificate you can share by link