Since 2020, Google has developed a vast ecosystem of tools aimed at analyzing website performance from the perspective of Core Web Vitals. Some rely on real user data, while others are based on on-demand test data. To fully understand what differentiates the PageSpeed Insights tool from the Lighthouse browser extension, it is essential to first examine these two families of tools.
What's the difference between field and lab data?
Field data is collected from your real visitors as they browse; lab data comes from a replay of a load in controlled conditions. The former measures what happens, the latter explains why. Neither replaces the other, and their numbers never match exactly.
There are two ways to study the performance level of a web page or site. Neither is better than the other: they complement each other to provide the most comprehensive overview of the User Experience possible.
What is field data?
Field data, or real-world data, is collected on your visitors' devices and connections, then aggregated at the 75th percentile. The public reference database is the Chrome User Experience Report, powered by Chrome users. This is the data Google uses to evaluate Core Web Vitals.
“Field data” is collected from real users during their web browsing. The only publicly accessible web performance database is the Chrome User Experience Report (CrUX), which consolidates the experience of Google Chrome browser users (multi-platform with the notable exception of iOS). Several conditions must be met, however, for the collection of this valuable information to be activated:
- be signed in to your Google account;
- have usage statistics sharing enabled;
- have history sync enabled;
- not have account data encryption enabled.
Many browser users thus share, often without even realizing it, a lot of information about how the sites they visit load. Consolidated within CrUX, this data is an invaluable source of information for Google. It is notably this data that is used as part of the Page Experience aspect, and in the Core Web Vitals tab of Google Search Console.
Given that this data is collected from multiple users, it is subject to high variability: depending on their location, connection type, hardware, or web browser extensions, two users will never have the same experience visiting a web page, and therefore the same performance metrics. This is why we work with averages/percentiles and the famous three-segment distribution gauge, which is very visual:
What is lab data?
Lab data, or synthetic data, comes from a load triggered on demand on a simulated machine and network. It is reproducible, immediately available, and detailed down to the line of code. It says nothing about your real visitors, but it alone allows for diagnosis.
Also known as synthetic data, they are collected on demand for the specific purpose of analyzing web performance. Their analysis can be launched from a user's terminal (via a browser, for example) or through dedicated server infrastructure, using crawlers. This is notably what audit tools like WebPageTest, GTmetrix, and their ilk offer.
In this second case, the goal is to ensure that the test conditions remain unchanged over time: same CPU and memory resources, same network connection (bandwidth and latency, etc.). This control over the technical environment allows for comparisons of the same site over time, or even between different sites, while being certain of having reliable and stable measurement indicators.
Unlike field data, the goal is to obtain unique performance metrics. They are thus expressed in units of time (FCP, LCP, Speed Index, Total Blocking Time...) or as an index (Cumulative Layout Shift), all associated with a color: green for "good," orange for "needs improvement," and red for "slow."
What do PageSpeed Insights and Lighthouse have in common?
Both tools run the same analysis engine, Lighthouse, and therefore produce the same performance score and recommendations for a given page. PageSpeed Insights is Lighthouse running on Google's servers, with imposed network and hardware configurations.
PageSpeed Insights and Lighthouse both offer the ability to run synthetic analyses instantly. The former will perform the test using Google's server architecture, while the latter will use your own browser and internet connection. The lab data thus retrieved is consolidated within a visually very similar interface, and for good reason: this component of PageSpeed Insights is based on Lighthouse. It is available for mobile and desktop.
This common layout is divided into three main sections:
- An overall performance score out of 100, calculated transparently from performance metrics and Core Web Vitals. You need to be above 50% to move from red to orange, and above 90% to be in the green.
- The main 5 performance statistics, including Largest Contentful Paint (LCP), Total Blocking Time (TBT), and Cumulative Layout Shift (CLS) metrics, which are intended to reflect the quality of the user experience as accurately as possible.
- A filmstrip view visually presenting the page loading steps. It allows for quick detection of issues such as excessively high server response times (TTFB), poor resource prioritization (LCP image, for example), or Layout Shifts.
Whether for PageSpeed Insights or Lighthouse, it is important to take a step back from this data: it only corresponds to a snapshot at a given moment and under conditions that likely do not reflect those of your real visitors. Having a good overall score and metrics in the green does not automatically guarantee that the site passes the Core Web Vitals in field data.
It is a testing tool that can be used to verify the impact of a deployment or to monitor a site in order to avoid regressions in web performance. But it should never, ever be considered a validation of a site's quality by Google: only field data collected in the Chrome User Experience Report counts.
Is your site as fast as your visitors expect?
What are the differences between PageSpeed Insights and Lighthouse?
PageSpeed Insights adds field data from CrUX, which Lighthouse lacks; Lighthouse, run locally, analyzes four sections instead of just one and accepts non-public pages. One is used to position a site, the other to diagnose it, including in pre-production or behind authentication.
| Criterion | PageSpeed Insights | Lighthouse |
|---|---|---|
| Field data | Yes, via CrUX | No |
| Lab data | Yes, Lighthouse engine | Yes, locally |
| Sections analyzed | Performance | Performance, accessibility, best practices, SEO |
| Execution environment | Google servers, imposed | Your machine, configurable |
| Private or local pages | No | Yes |
| Recommended use | Position and track | Diagnose and test |
The decisive line in this table is the first one. Only PageSpeed Insights exposes field data, the data Google uses to evaluate your site: a lab score of 100 on a page with poor field Core Web Vitals is worthless. The reverse is also true, and much more common than one might think.
Beyond their synthetic performance analysis component, PageSpeed Insights and Lighthouse each offer a complementary component with high added value. Both are useful for in-depth technical analysis of a site.
Lighthouse: Beyond Performance
Lighthouse isn't limited to performance: it also analyzes accessibility, best practices, and SEO, each scored out of 100. These three sections are absent from the public PageSpeed Insights score. This makes it a testing tool, executable before each production release.
While the Lighthouse extension for Google Chrome and Firefox is a very popular tool for studying a page's web performance, it is not limited to this category of analysis. Four sections can be activated as needed to provide a more or less exhaustive view of a site's technical status:
- Performance: Is the site fast to load and usable under good conditions?
- Accessibility: Is the page accessible to the widest possible range of users?
- Best Practices: Is the page developed in compliance with HTML, CSS, and JavaScript standards?
- SEO: Does the page adopt basic on-site optimizations for natural referencing?
As web performance is a specialized technical discipline, it is common for experts to address topics such as accessibility, best practices, or SEO. Therefore, Lighthouse's complementary components can sometimes be useful for taking the temperature on these subjects. This is a first approach that can only enrich the understanding of a site's technical issues.
PageSpeed Insights: Field Data Included
PageSpeed Insights displays, above the lab score, the Core Web Vitals actually measured on your site by Chrome users. This section only appears if your traffic is sufficient to feed CrUX. Its absence signals a site that is too little visited to be included in the public database.
As we saw previously, lab data alone cannot and should not be sufficient to judge the performance of a website. This is why PageSpeed Insights includes a large section dedicated to field data, and why this section is presented first and foremost, even before the results of the synthetic analysis. It is therefore important to clearly distinguish between the "Discover your users' experience" section (field data) and the one labeled "Diagnose performance issues" (lab data).
As this tool is particularly rich and interesting, we will break down the sections that make it up together:
- To the right of the title, a toggle allows you to switch between metrics specific to the page (This URL) and those for the entire site (Origin). The Origin view can be very useful for detecting transversal web performance issues.
- At the top center, it is indicated whether the page or site passes the Core Web Vitals assessment. In our screenshot, the page is failing.
- At the center are the performance metrics, including the famous Core Web Vitals. Since this is field data, the bar chart presentation introduced earlier is used. At a glance, we can see here that LCP is the issue for the majority of visitors, followed by FCP and INP.
- The last section on a gray background details the conditions under which the data was collected. One of the important pieces of information it contains is the mention of the collection period: rolling 28 days. Core Web Vitals have a form of inertia (some would say slowness): they are not updated in real time because it would require too many server resources from Google.
How to interpret a Lighthouse score?
A Lighthouse score is read in three tiers: from 90 to 100 the score is good and displayed in green, from 50 to 89 it requires fixes and is displayed in orange, below 50 it is bad and displayed in red. This score is not a measure of site quality, but the weighted aggregation of five lab metrics.
Weighting explains most misunderstandings. The performance score does not treat the five metrics equally: some weigh three times more than others in the calculation. Gaining ten points does not require the same effort depending on the target metric, and two sites with the same score can have opposite profiles.
Second source of confusion: variability. Two successive runs on the same page commonly yield scores that differ by several points, depending on the load of the network, your machine, and current third-party scripts. A difference of less than five points is not a signal, it's measurement noise.
Finally, the lab score has no ranking value. Google does not use the Lighthouse score as a ranking criterion: it uses the Core Web Vitals measured in the field, at the 75th percentile. Aiming for 100 on Lighthouse is a means, never the objective, and a site can very well pass the Core Web Vitals with a score of 75.
Which tool to choose between PageSpeed Insights and Lighthouse?
Use PageSpeed Insights to know where you stand, Lighthouse to understand why and fix it. The first is the arbiter, as it carries the data Google uses; the second diagnoses, on any page, even private ones. Relying solely on the lab score is the most common mistake.
As you will have understood, both tools allow you to approach web performance from a particular angle. We cannot recommend one over the other, and for good reason: they complement each other at all levels. Better yet, they should ideally only constitute one component of your ecosystem of tools for testing and monitoring.
At Agence Web Performance, where we perform daily web performance audits, we use other tools in addition to those developed by Google's teams. This is the case for Webpagetest and GTmetrix, which offer many more and advanced possibilities in terms of testing options. And dozens of other more specific complementary tools, either focused on a particular metric (CLS in particular), or on certain advanced aspects (security, http headers, fonts...).
In absolute terms, the only data you cannot do without is that of the Chrome User Experience Report. However, if you have enabled Google Search Console, you already have access to it via the Core Web Vitals tab with much more detail than in PageSpeed Insights. Neither PageSpeed Insights nor Lighthouse are thus indispensable: the essential thing is to ensure that your site is fast, perhaps by relying on the expertise of a specialist like Agence Web Performance 😉
