Why Lighthouse Scores Can Vary Between Runs
Website intelligenceWhy Lighthouse Scores Can Vary Between Runs
A performance score that changes between runs does not automatically mean the website has changed. Browser-based testing is affected by the environment, so comparisons need consistent context.
Recognize the variables in a lab test
Network conditions, browser workload, CPU availability, caching, third-party scripts and server responses can all influence a run. A page can produce slightly different measurements even when its code has not changed.
A large score movement deserves investigation, but first check whether the test used the same URL, device profile and audit source.
Compare the same report context
When SiteBoost displays Lighthouse results, note the scan date, URL, source and status. Compare like with like rather than treating a scan from a different page or environment as a direct before-and-after measurement.
If the result is surprising, repeat the test and inspect the underlying metrics. The summary score should lead you toward the detail, not replace it.
- Use the same public URL.
- Keep the test device and network context consistent where possible.
- Compare the same Lighthouse category and source.
- Look for a repeated pattern before making a broad change.
Document the change that occurred
Record which deployment, asset or configuration changed between tests. That makes it easier to connect a real improvement or regression to the implementation.
Do not optimize for a single isolated number when the visitor experience and underlying metrics tell a different story.
