Performance testing is a type of non-functional testing that measures how a system behaves under a given workload. Rather than checking whether a feature works (that is functional testing), performance testing asks a different question: how fast, how stable, and how scalable is the system when real users put it under pressure?
A feature can be perfectly correct and still fail in production if it is slow, times out under load, or crashes when too many users arrive at once. Performance testing exists to catch those problems before customers do.
Why performance testing matters
- User experience — slow pages drive users away; speed directly affects satisfaction and conversions.
- Reliability — it reveals whether the system stays stable during peak traffic.
- Scalability — it shows whether the system can grow with demand.
- Cost and capacity planning — it informs how much infrastructure you actually need.
- Risk reduction — finding bottlenecks early is far cheaper than fixing outages in production.
Key metrics
Performance testing is driven by measurable numbers. The most important ones include:
- Response time — how long a request takes, often reported at averages and percentiles (e.g. 90th/95th percentile).
- Throughput — how many requests or transactions the system handles per second.
- Concurrency / virtual users — how many simultaneous users are simulated.
- Error rate — the percentage of failed requests under load.
- Resource utilisation — CPU, memory, disk, and network usage on the servers.
- Latency — the delay before a transfer of data begins.
Where it fits
Performance testing is usually done after functional testing has confirmed the application behaves correctly, and ideally in an environment that closely mirrors production. It is a specialised discipline that combines test design, scripting, load generation, and careful analysis of results.
Conclusion
Performance testing answers the questions functional testing cannot: is the system fast, stable, and scalable enough for real-world use? By measuring response time, throughput, error rates, and resource usage under realistic load, teams can deliver software that performs well when it matters most. The following articles in this series dive deeper into tools, strategy, test types, and modern challenges.