Overview
It Worked Yesterday
You cannot detect a regression if you don't know what 'normal' looks like. The baseline is your anchor in the chaos of continuous deployment.
In QA, baselines are most critical in Performance Testing and Visual Regression Testing. If your login page takes 2 seconds to load today, is that good? If your baseline says it used to take 0.5 seconds, then no, it's a failure. Establishing a baseline prevents the 'boiling frog' syndrome where performance degrades slowly over time without anyone noticing.
Best Practices
Dos and Don'ts
Avoid common mistakes that can lead to flaky tests and maintenance nightmares.
What to do
- •Establish baselines early in the project.
- •Store baseline data in a version-controlled repository.
- •Update the baseline only when a change is intentional and verified.
Common Pitfalls
- •Don't compare apples to oranges (e.g., Staging baseline vs. Production results).
- •Don't ignore deviations just because they are small.
- •Don't lose the historical context of previous baselines.
The Details
Visual Baselines
In Visual Testing (using tools like Percy or Applitools), the baseline is a screenshot of the UI that has been approved by a human. Every subsequent test run compares the new screenshot pixel-by-pixel against this baseline. The 'Diff' is what the QA reviews.