PageSpeed is a tool from Google that analyses how well a webpage loads on both desktop and mobile devices. It provides a performance score, along with practical recommendations to help improve the user experience.
While the score itself is not a direct Google ranking factor, many of the metrics it measures form part of Google’s Core Web Vitals, which are used as signals within Google’s search ranking systems.
PageSpeed scores are a useful indicator, but they need to be interpreted carefully. They are not simply a measure of how quickly a website loads, but rather how efficiently it loads from a technical perspective. For example, PageSpeed assesses how files are prioritised, how much processing the web browser has to perform, and whether features such as tracking scripts, cookie consent tools, web fonts, and anti-spam checks delay the page from becoming visible and interactive.
This distinction is important because a low PageSpeed score does not necessarily mean visitors are experiencing a slow website. A page may achieve a relatively modest score while still feeling fast and responsive to users, particularly if the content they need appears quickly.
Real user data vs simulated testing
The “real user” data shown in PageSpeed reports is based on the experience of actual visitors to your website. It reflects real devices, real network conditions, and real browsing behaviour over the previous 28 days.
The simulated PageSpeed test, however, can produce significantly slower results (sometimes in excess of ten seconds). This is because it measures performance under a controlled set of conditions designed to represent a lower-powered mobile device. The test deliberately applies reduced processing power, slower network speeds, and an empty browser cache to help identify potential performance bottlenecks.
These test conditions can make the impact of JavaScript and third-party services appear much greater than they would for many real visitors. As a result, websites with interactive features or integrations may receive lower simulated scores, even though they feel fast and responsive in everyday use.
A typical website may include a surprisingly large number of third-party services, such as Google Analytics, Facebook tracking, cookie consent tools, embedded YouTube videos, anti-spam services (such as reCaptcha), and web font providers. Each adds a small amount of processing, which PageSpeed includes as part of its assessment.
Why real users often see better performance
Real visitors often experience faster loading times than PageSpeed’s simulated test because they benefit from factors such as:
- Cached files from previous visits, allowing some content to load instantly.
- Faster devices than the emulated mobile device used by PageSpeed.
- More efficient loading of resources, with multiple files downloading and processing in parallel.
- Faster network connections than those used in the simulated test.
This is why real-user performance data is often significantly better than simulated test results.
Why PageSpeed still matters?
Although PageSpeed should not be treated as a direct indicator of user-perceived loading speed, it remains an important diagnostic tool and a valuable SEO metric.
It helps identify technical inefficiencies that may not be obvious during normal browsing, such as render-blocking resources, excessive JavaScript, or opportunities to improve loading priorities. These issues can become more noticeable as additional functionality is added to a website over time.
Why do PageSpeed scores fluctuate?
It is perfectly normal for PageSpeed scores to vary between tests, which is why we typically run multiple tests and look for overall trends rather than focusing on a single result.
Scores can fluctuate for a number of reasons, including:
- Recent changes to the website, which may clear and rebuild cached content.
- Variations in the response times of third-party services.
- Temporary network conditions during the test.
- Minor differences in the test environment from one run to the next.
For this reason, individual PageSpeed results should always be viewed in context rather than in isolation.
Why performance can improve… but still look poor?
One quirk of PageSpeed is that improvements are not always reflected proportionally in the overall score. It is possible to resolve one performance issue, only for PageSpeed to place greater emphasis on another bottleneck, resulting in a similar or only modestly improved score. This can make progress seem less visible, even when the underlying performance has genuinely improved.
Third-party services can have a particularly noticeable impact on simulated testing because they introduce additional requests, external dependencies, and processing overhead. As websites become more feature-rich, with services such as analytics, cookie consent tools, embedded content, and marketing integrations, these cumulative overheads can make it more difficult to achieve very high PageSpeed scores.
For this reason, PageSpeed is best viewed as a diagnostic tool for identifying optimisation opportunities, rather than as a literal measure of how many seconds a visitor waits for a page to load. The most meaningful assessment combines PageSpeed testing with real-user performance data and practical testing on real devices.