Why your website speed score varies between tests

If you have tested your website speed more than once, you have probably noticed the score changes between tests. This is normal and expected. This guide explains why it happens and what to focus on instead of chasing a single number.

Why scores differ between tools

Different speed testing tools use different methods to measure your site. The three most common tools are GTmetrix, Google PageSpeed Insights, and the Web60 speed test. Each one produces different scores for the same website because of how they work.

Test server location

GTmetrix runs tests from various locations around the world. If the test server is in Canada and your website is hosted in Dublin, the physical distance adds latency. Google PageSpeed Insights uses its own global infrastructure. The Web60 speed test runs from the same Dublin data centre where your site is hosted, which gives the most representative result for visitors in Ireland.

Device simulation

Each tool simulates different devices. Google PageSpeed Insights simulates a mid-range Android phone on a 4G connection for its mobile test. GTmetrix uses its own device profiles. The simulated device speed and network throttling directly affect the score.

Scoring methodology

The tools weight their metrics differently. One tool might penalise a slow Largest Contentful Paint more heavily than another. The same raw performance data can produce a score of 75 on one tool and 85 on another simply because of how each tool calculates the final number.

Why scores differ between runs on the same tool

Even running the same tool twice in a row can produce different scores. Several factors cause this:

• Cache state. The first test after a cache clear will be slower because the server must build the page fresh. The second test hits the cached version, which is faster.
• Server load. Minor fluctuations in server load between tests can affect response times by a few milliseconds, which influences the score.
• Network conditions. The internet path between the test server and your website changes subtly with every request. Small differences in routing or congestion affect timing.
• Third-party scripts. If your site loads scripts from external services (analytics, fonts, chat widgets), the response time of those external servers varies between tests.

Fluctuations of 5 to 10 points between consecutive tests are completely normal.

What to focus on instead

Rather than reacting to a single test result, track your scores over time and look for trends:

• Consistent green range (90+): Your site is performing well. No action needed.
• Consistent amber range (50-89): There is room for improvement, but your site is functional. Review images and plugins.
• Consistent red range (below 50): There are likely specific issues worth addressing. Check for oversized images, excessive plugins, or caching problems.
• Sudden drop from green to red: Something changed. Review any recent plugin installations, theme changes, or large images you added.

A score that stays consistently in the 80 to 90 range is more meaningful than a single test that hit 95. Focus on keeping your baseline healthy rather than optimising for one perfect score.

The Web60 speed test

The speed test in the Web60 portal runs from the same Dublin data centre as your site. This makes it the most consistent and relevant test for your visitors, particularly if your audience is primarily in Ireland and the UK. Use it as your primary benchmark and treat other tools as supplementary data.