What is Fake Data?

Fake data consists of structurally plausible attributes assembled on the fly from localized wordlists and seeded pseudorandom number generators (PRNGs). When you use a fake user generator, it constructs an artificial profile with a first name, surname, street address, and an email address pointing to a reserved domain like example.com per IETF RFC 2606.

There is no underlying statistical model connecting the fields: a fake record might assign a senior engineering executive job title to an 18-year-old profile. For UI component testing, CSS responsive layout checks, and form validation, that lack of statistical correlation is completely irrelevant. What matters is catching edge cases like 40-character names, non-ASCII characters, and telephone formatting bugs.

• Primary purpose: UI styling, form validation tests, database seeding, edge-case checking (e.g. non-ASCII names, long text strings).
• Risk profile: Zero PII (Personally Identifiable Information) risk when using verified reserved ranges (such as SSA 900-series unissued SSNs).

What is Mock Data?

Mock data is static, predetermined fixture data specifically crafted to force a specific code execution path. In unit tests and integration tests, you don't want randomness; you want certainty. If you are testing how your checkout screen handles an expired card, you don't want a randomized expiration date; you want a fixed fixture that triggers an HTTP 402 Payment Required response every time.

Mock data lives in your __fixtures__ directory or inside your Mock Service Worker (MSW) handlers. It is minimal, rigid, and fast.

• Primary purpose: Fast unit tests, contract testing, continuous integration (CI) test runners, and decoupled frontend development.
• Risk profile: Zero risk, provided developers do not copy-paste raw payloads containing bearer tokens or active customer records from production logs.

What is Synthetic Data?

Synthetic data is mathematically generated data designed to preserve the mathematical distribution, covariance, and statistical properties of an authentic production dataset. Unlike fake data (which draws independently from static dictionary lists), synthetic data engines use machine learning models (such as CTGANs, Copula models, or Bayesian networks) to ensure that if older users in your real database have higher transaction values, the synthetic output exhibits the exact same correlation.

Data science teams, healthcare researchers, and database query optimization engineers rely on synthetic datasets to train models and benchmark query planners without exposing real patient or customer identities.

• Primary purpose: ML model training, database query planner optimization, load and stress testing, data warehousing pipelines.
• Risk profile: Extremely low, but requires mathematical privacy bounds ($\epsilon$-differential privacy) to prevent re-identification attacks.

json-server creates a fully operational, authentic REST API with sorting, pagination, and persistent CRUD operations in 30 seconds.

1. Generate 100 fake user records on the Fake User Generator.
2. Click Export → JSON and save the file as users.json.
3. Wrap the array inside an object in a file named db.json:

{
  "users": [
    {
      "id": 1,
      "name": "Jane Smith",
      "email": "jane.smith@example.com",
      "city": "Springfield",
      "job": "Frontend Engineer"
    }
  ]
}

Start the local mock server:

npx json-server --watch db.json --port 3001

You now have working endpoints: GET /users?_page=1&_limit=10, POST /users, PUT /users/1, and DELETE /users/1.

Mock Service Worker intercepts requests at the browser network layer using Service Workers. Your React, Next.js, or Vue components call normal production URLs like fetch('/api/v1/users') without knowing they are mocked.

import { http, HttpResponse } from 'msw';
import fakeUsers from './fixtures/fake-users.json';

export const handlers = [
  // Intercept GET requests to /api/users
  http.get('/api/users', ({ request }) => {
    const url = new URL(request.url);
    const limit = Number(url.searchParams.get('limit')) || 10;
    return HttpResponse.json(fakeUsers.slice(0, limit), { status: 200 });
  }),

  // Intercept POST requests to add a user
  http.post('/api/users', async ({ request }) => {
    const newUser = await request.json();
    return HttpResponse.json({ id: Date.now(), ...newUser }, { status: 201 });
  })
];

If your QA team relies on Postman collections, export a dataset from our tool and save it as an example response for your request. Postman will generate a public or private mock server URL that responds with your exact JSON payload and matches specified query parameters.

In mobile applications (such as Flutter using Dio or React Native using axios), add an interceptor that checks for a debug environment flag. When enabled, the interceptor short-circuits the HTTP request and returns bundled mock assets, allowing full offline UI development on physical simulators and staging devices.

Live API mode

This mode makes a direct client-side request from your browser to the public randomuser.me endpoint. That service assembles records from large pools of invented names, surnames, street names, cities and email domains, and attaches photographs of synthetic people. Because the photographs are real image files rather than SVG placeholders, this mode is ideal for visual components: avatar stacks, profile cards, comment threads, and responsive tables with image columns.

The upstream service is a convenience rather than a dependency: requests above 100 records are filled in locally, and if the endpoint is unreachable, the entire batch seamlessly fails over to offline generation.

Offline deterministic mode

This mode never touches the network. Names, streets, cities, job titles, and company names are assembled from dictionaries compiled directly into the page, combined with a mulberry32 seeded pseudo-random number generator. Because the generator is deterministic, entering the same seed twice produces exactly the same dataset. That matters for automated CI/CD test runners: a test fixture you cannot reproduce is a bug you cannot chase.

Offline mode also generates developer-specific fields including job title, company size, registration date, UUID v4, IPv4 address, MAC address, and secure password hashes.