Test Data Strategy

When to Use This Skill

Use this skill when:

• Test Data Strategy tasks - Working on plan comprehensive test data management including synthetic data generation, data anonymization, versioning, and environment-specific strategies
• Planning or design - Need guidance on Test Data Strategy approaches
• Best practices - Want to follow established patterns and standards

Overview

Effective test data management ensures tests have the right data at the right time while protecting sensitive information and maintaining data quality across environments.

Test Data Types

Type	Source	Use Case	Privacy Risk
Synthetic	Generated	Unit/Integration tests	None
Subset	Production sample	Performance testing	Medium
Masked	Anonymized production	Realistic scenarios	Low
Production Clone	Full copy	Pre-prod validation	High
Baseline	Curated reference	Regression testing	Low

Test Data Strategy Template

# Test Data Strategy: [Project Name]

## 1. Data Requirements

### By Test Level
| Level | Data Source | Volume | Refresh |
|-------|-------------|--------|---------|
| Unit | Synthetic | Minimal | On-demand |
| Integration | Synthetic/Subset | Moderate | Per run |
| System | Masked production | Realistic | Weekly |
| Performance | Scaled synthetic | Production-like | Per release |

### By Feature Area
| Feature | Critical Data | Volume Required | Sensitivity |
|---------|---------------|-----------------|-------------|
| Authentication | User accounts | 1000 | High |
| Payments | Transactions | 10000 | High |
| Reporting | Historical data | 1M records | Medium |

## 2. Data Generation Strategy

### Synthetic Data Tools
- **Unit Tests**: AutoFixture, Bogus
- **Integration**: TestContainers + Seed
- **Performance**: Bulk generators

### Generation Rules
| Entity | Key Fields | Generation Logic |
|--------|------------|------------------|
| User | Email | `{guid}@test.example.com` |
| Order | Amount | `Random(1, 10000)` |
| Date | Timestamp | `Random(now-1y, now)` |

## 3. Data Anonymization

### PII Fields
| Field | Original | Anonymization Method |
|-------|----------|---------------------|
| Name | John Smith | Faker generated |
| Email | john@acme.com | `hash@domain.test` |
| Phone | 555-123-4567 | `555-xxx-xxxx` |
| SSN | 123-45-6789 | `xxx-xx-xxxx` |
| Address | 123 Main St | Faker address |
| DOB | 1985-03-15 | Shift by random days |

### Anonymization Rules
- Preserve data relationships
- Maintain referential integrity
- Keep statistical properties
- Remove unique identifiers

## 4. Environment Strategy

### Dev Environment
- Source: 100% synthetic
- Refresh: On-demand
- Volume: Minimal

### QA Environment
- Source: Masked production subset
- Refresh: Weekly
- Volume: 10% of production

### Staging Environment
- Source: Masked production clone
- Refresh: Before each release
- Volume: 100% of production

### Performance Environment
- Source: Scaled synthetic
- Refresh: Before performance runs
- Volume: 150% of production

## 5. Data Versioning

### Baseline Management
- Version baseline data sets
- Track data schema changes
- Maintain backward compatibility
- Document data dependencies

### Refresh Procedures
1. Trigger: [Manual/Scheduled/Event]
2. Source: [Production/Backup/Generator]
3. Transform: [Anonymization steps]
4. Load: [Target environment]
5. Validate: [Verification checks]

## 6. Compliance Requirements

### GDPR Compliance
- [ ] No real EU citizen data in non-prod
- [ ] Right to erasure supported
- [ ] Data minimization applied
- [ ] Consent tracking anonymized

### HIPAA Compliance
- [ ] PHI fully de-identified
- [ ] Safe Harbor method applied
- [ ] Audit logs maintained
- [ ] Access controls verified

Synthetic Data Generation (.NET)

Using Bogus

using Bogus;

public class TestDataGenerator
{
    public static Faker<Customer> CustomerFaker => new Faker<Customer>()
        .RuleFor(c => c.Id, f => f.Random.Guid())
        .RuleFor(c => c.FirstName, f => f.Person.FirstName)
        .RuleFor(c => c.LastName, f => f.Person.LastName)
        .RuleFor(c => c.Email, (f, c) => f.Internet.Email(c.FirstName, c.LastName))
        .RuleFor(c => c.Phone, f => f.Phone.PhoneNumber())
        .RuleFor(c => c.DateOfBirth, f => f.Date.Past(50, DateTime.Now.AddYears(-18)))
        .RuleFor(c => c.Address, f => new Address
        {
            Street = f.Address.StreetAddress(),
            City = f.Address.City(),
            State = f.Address.StateAbbr(),
            Zip = f.Address.ZipCode()
        });

    public static Faker<Order> OrderFaker(Customer customer) => new Faker<Order>()
        .RuleFor(o => o.Id, f => f.Random.Guid())
        .RuleFor(o => o.CustomerId, customer.Id)
        .RuleFor(o => o.OrderDate, f => f.Date.Recent(30))
        .RuleFor(o => o.Total, f => f.Finance.Amount(10, 1000))
        .RuleFor(o => o.Status, f => f.PickRandom<OrderStatus>());
}

Using AutoFixture

using AutoFixture;
using AutoFixture.Xunit2;

public class CustomerTests
{
    [Theory, AutoData]
    public void CreateCustomer_WithValidData_Succeeds(Customer customer)
    {
        // AutoFixture generates valid Customer automatically
        var result = _service.Create(customer);
        Assert.True(result.IsSuccess);
    }

    [Theory, AutoData]
    public void ProcessOrder_CalculatesCorrectTotal(
        [Frozen] Customer customer,
        Order order,
        List<OrderItem> items)
    {
        // Frozen ensures customer is reused
        // Order and items are auto-generated
        order.Items = items;
        var total = _calculator.Calculate(order);
        Assert.Equal(items.Sum(i => i.Quantity * i.Price), total);
    }
}

Seeding Test Databases

public class TestDatabaseSeeder
{
    public static async Task SeedAsync(AppDbContext context)
    {
        // Clear existing data
        await context.Database.ExecuteSqlRawAsync("DELETE FROM Orders");
        await context.Database.ExecuteSqlRawAsync("DELETE FROM Customers");

        // Generate test data
        var customers = TestDataGenerator.CustomerFaker.Generate(100);
        await context.Customers.AddRangeAsync(customers);

        foreach (var customer in customers)
        {
            var orders = TestDataGenerator.OrderFaker(customer).Generate(5);
            await context.Orders.AddRangeAsync(orders);
        }

        await context.SaveChangesAsync();
    }
}

Data Anonymization Techniques

Technique	Description	Use Case
Substitution	Replace with fake data	Names, emails
Shuffling	Rearrange within column	Salaries, dates
Masking	Partial hiding	SSN (xxx-xx-1234)
Generalization	Reduce precision	Age ranges, zip prefix
Nulling	Remove entirely	Unnecessary fields
Tokenization	Replace with token	Cross-reference needs
Hashing	One-way transform	Identifiers

.NET Anonymization Example

public class DataAnonymizer
{
    public Customer Anonymize(Customer source)
    {
        return new Customer
        {
            Id = source.Id, // Preserve for relationships
            FirstName = _faker.Person.FirstName,
            LastName = _faker.Person.LastName,
            Email = $"{Guid.NewGuid():N}@test.example.com",
            Phone = MaskPhone(source.Phone),
            SSN = "xxx-xx-" + source.SSN.Substring(7, 4),
            DateOfBirth = ShiftDate(source.DateOfBirth),
            Address = new Address
            {
                Street = _faker.Address.StreetAddress(),
                City = source.Address.City, // Preserve geography
                State = source.Address.State,
                Zip = source.Address.Zip.Substring(0, 3) + "00"
            }
        };
    }

    private string MaskPhone(string phone)
    {
        // Keep area code, mask rest
        return Regex.Replace(phone, @"(\d{3})\d{3}(\d{4})", "$1-xxx-$2");
    }

    private DateTime ShiftDate(DateTime date)
    {
        // Shift by random days within ±30
        return date.AddDays(_random.Next(-30, 30));
    }
}

Test Data Patterns

Builder Pattern

public class CustomerBuilder
{
    private Customer _customer = new();

    public CustomerBuilder WithName(string first, string last)
    {
        _customer.FirstName = first;
        _customer.LastName = last;
        return this;
    }

    public CustomerBuilder WithPremiumStatus()
    {
        _customer.IsPremium = true;
        _customer.PremiumSince = DateTime.Now.AddYears(-1);
        return this;
    }

    public CustomerBuilder WithOrders(int count)
    {
        _customer.Orders = TestDataGenerator.OrderFaker(_customer).Generate(count);
        return this;
    }

    public Customer Build() => _customer;
}

// Usage
var customer = new CustomerBuilder()
    .WithName("Test", "User")
    .WithPremiumStatus()
    .WithOrders(5)
    .Build();

Object Mother Pattern

public static class TestCustomers
{
    public static Customer ValidCustomer() => new()
    {
        Id = Guid.NewGuid(),
        FirstName = "Test",
        LastName = "User",
        Email = "test@example.com",
        Status = CustomerStatus.Active
    };

    public static Customer PremiumCustomer() => new()
    {
        Id = Guid.NewGuid(),
        FirstName = "Premium",
        LastName = "User",
        Email = "premium@example.com",
        IsPremium = true,
        Status = CustomerStatus.Active
    };

    public static Customer InactiveCustomer() => new()
    {
        Id = Guid.NewGuid(),
        Status = CustomerStatus.Inactive
    };
}

Integration Points

Inputs from:

• Data model → Test data structure
• Privacy requirements → Anonymization rules
• test-strategy-planning skill → Data volume needs

Outputs to:

• Test automation → Data fixtures
• performance-test-planning skill → Load data
• Environment provisioning → Seed scripts