Published
9 MIN READ
PageSpeed Score: What It Proves About Your Website
The PageSpeed score is the result of a lab test: Lighthouse loads the page on a simulated phone and returns a number from 0 to 100. It's useful for finding problems, but it doesn't measure the experience of the people who visit the website. For search, Google uses Core Web Vitals measured on real visitors.
In Brazil, the difference matters. In August 2026, only 33.9% of websites measured with visitors in Brazil passed all three Core Web Vitals on mobile, versus 53.0% worldwide, according to the HTTP Archive Core Web Vitals Technology Report. Meanwhile, agencies at the top of search results promise a guaranteed score of 90, and the most-read guides in Portuguese teach how to raise the score without showing this data.
At Wama, websites are delivered in Framer, and after delivery the client only pays for the Framer subscription that stays in the client's account, with hosting, SSL, CDN and security updates included. The platform takes care of the server; everything else applies to any website.
What does the PageSpeed score measure?
The score measures a single page load under controlled conditions. According to the PageSpeed Insights documentation, Google runs Lighthouse simulating a mid-range phone on a mobile network and rates 90 to 100 as good, 50 to 89 as needs improvement and below 50 as poor.
In Lighthouse version 10, the score weighs five metrics: total blocking time (30%), LCP, the largest visible element (25%), CLS, layout stability (25%), and two first-paint metrics (10% each), according to the Lighthouse performance scoring explainer.
Each metric is turned into a score using a curve drawn from real HTTP Archive websites: the time only the fastest 8% achieve is worth 90, and the time of the fastest 25% is worth 50. A score of 90 means that load ranked among the fastest on the web, on that simulated device, at that moment. In Brazil, the median lab score on mobile is around 50.
And the moment changes. Framer's help article on PageSpeed Insights, updated on August 7, 2026, records variations of up to 15 points between tests of the same page and recommends running at least three times and using the median.
Lab vs. field data: what Google uses in search
For search, field data is what counts. Core Web Vitals come from the Chrome User Experience Report (CrUX), which aggregates real visits from the last 28 days and looks at the 75th percentile. A page passes when LCP is at most 2.5 seconds, INP at most 200 milliseconds and CLS at most 0.1, all three at the same time.
In the PageSpeed Insights FAQ, Google writes that good lab data doesn't necessarily mean a good real-world experience. On its page about page experience in Search results, updated in December 2025, it says Core Web Vitals are part of ranking, but that a good result doesn't guarantee the top spot.
Website speed sits alongside content, security and structure, without replacing them, and it's worth checking what to require at delivery so search engines and AI can find the website.
How many websites pass Core Web Vitals in Brazil?
A little more than one in three. In the August 2026 edition of the HTTP Archive report, which crosses CrUX with the technology detected on each website, 33.9% of origins measured with visitors in Brazil passed all three metrics on mobile. The country's bottleneck is loading: only 48.3% had good LCP.
Platform | Passing on mobile, visitors in Brazil | Passing on mobile, worldwide | Median lab score on mobile | Origins measured in Brazil |
|---|---|---|---|---|
All websites | 33.9% | 53.0% | 50 | 604,494 |
WordPress | 33.4% | 48.7% | 48 | 142,301 |
WordPress with Elementor | 27.9% | 36.8% | 45 | 79,890 |
Framer | 58.3% | 70.6% | 67 | 1,068 |
Wix | 68.1% | 81.2% | 61 | 13,841 |
Webflow | 45.1% | 71.2% | 56 | 3,420 |
Shopify | 51.6% | 76.5% | 48 | 8,699 |
Squarespace | 28.3% | 72.1% | 28 | 644 |
Hostinger Website Builder | 23.4% | 54.7% | 62 | 1,262 |
Duda | 60.4% | 86.0% | 46 | 1,494 |
GoDaddy Website Builder | 79.7% | 83.8% | 62 | 227 |
WordPress, the largest base measured in Brazil, sits at 33.4%, right on the country average, and drops to 27.9% with the Elementor builder. Framer passes at 58.3% here. Wix, Duda and the GoDaddy builder pass more often than it does, the last one with only 227 origins. In twelve months, the country went from 31.5% to 33.9%.
Wama has a stated position against WordPress for company websites, because of plugin vulnerability risk, permanent maintenance and the three-year cost. Clients who migrated from WordPress to the studio were spending US$ 60 to US$ 200 per month on hosting, plugins and maintenance combined, which is the math behind how much hosting and plugins cost per month.
What the table doesn't prove
The report assigns technology based on the homepage, only includes websites with enough Chrome traffic, and "Brazil" means visitors in Brazil, identified by IP, according to the CrUX dimensions documentation. The HTTP Archive methodology post warns that cause and effect can't be determined; each platform brings together different kinds of websites.
A fast website is one criterion among many, and choosing between Framer, WordPress and Webflow on other criteria weighs editing, cost and maintenance.
When the score and the field data disagree
Three hosted builders have a median lab score between 61 and 62 on mobile, and their pass rate with visitors in Brazil is 23.4% for Hostinger's, 68.1% for Wix and 79.7% for GoDaddy's. Worldwide, Squarespace has a median score of 28 and 72.1% of origins passing.
The lab test uses one device, one network and one location; field data combines phones, connections, cache and the behavior of real people, as web.dev describes in why lab and field data can be different. A website can do poorly in the simulated test and well in the visitor's pocket, and vice versa.
In Framer, the weak spot in Brazil is INP, the response to a tap: 62.6% of origins have good INP, versus 75.1% with good LCP. The web.dev guide to optimizing INP, updated in September 2025, explains that heavy scripts create long tasks that delay the response, a cost that grows with every third-party tool installed.
If your website has shown poor field data for months, the question stops being the score and becomes when fixing the current website stops being worth it.
How do you check an agency's PageSpeed 90 promise?
Check the field data of websites the agency has already published, not the score of a demo page. It takes a few minutes, with the same tool the agency cites.
Ask for three portfolio URLs that have been live for more than 28 days and get visits.
Open each one in PageSpeed Insights, on the mobile tab, and read the real-user experience section first.
If that section says there isn't enough data, the website has little traffic, and the lab score alone doesn't prove real experience.
Run the lab test three times and compare the median, never a single run.
For your own company's website, use the Core Web Vitals report in Search Console.
Wama's portfolio is open on the website, with more than 200 projects delivered and more than 50 published in Framer: any URL there can go through the same test before a sales conversation. Bring along the other questions that fit in the same conversation with the agency.
What to write in the proposal instead of a score of 90
A guaranteed score in a proposal is worth little, because it fluctuates and is measured in a scenario your visitor doesn't reproduce. What you can hold the vendor to are scope decisions that affect field data, written down before the design:
a weight limit for images and videos on the homepage;
a list of approved third-party scripts (chat, pixel, tag manager) and who authorizes a new one;
a review of field data after 28 days of traffic, including what happens if mobile doesn't pass;
who answers when marketing installs a new tool and INP gets worse.
At Wama, custom animation and interaction is one of the three items that move a website budget, alongside the number of pages and who writes the copy. Visual effects are a scope decision and need to be in writing, like everything that defines what goes into the price of a website project.
Anyone leaving WordPress because of performance first needs to move a website from WordPress to Framer while keeping the URLs, so they don't trade slowness for a drop in traffic.
Frequently asked questions about the PageSpeed score
Does a PageSpeed score of 100 improve your Google ranking?
Not directly. Google uses Core Web Vitals measured on real visitors and states that a good result on them doesn't guarantee the top of search. A score of 100 is a lab result and can coexist with poor field data. Passing all three metrics on mobile is worth more than the last point on the test.
Why does the PageSpeed score change with every test?
Because the test depends on the network, ads, A/B tests, browser extensions and the machine running Lighthouse. Chrome's documentation lists these sources of variability, and Framer support has recorded differences of up to 15 points on the same page. Run it three times and compare the median, never a single run.
What should you do when PageSpeed says there's no real-user data?
It means the page or website doesn't get enough Chrome visits to be included in CrUX. In that case, the lab score is the only number available and it's useful for finding problems, not for proving experience. Field data appears once traffic grows, and then it's worth checking again.
PageSpeed Insights or Search Console: which should you use?
Both. PageSpeed Insights tests one URL at a time and shows lab and field data side by side, which is good for checking an agency's portfolio. The Search Console Core Web Vitals report shows how the pages on your own website perform with real visitors, and it's only available to people with verified access.
Do Framer websites pass Core Web Vitals?
Most of the ones measured do. In August 2026, 58.3% of Framer origins passed on mobile with visitors in Brazil, according to HTTP Archive, above the country average. The platform helps, but the result depends on what goes into the page: images, videos and third-party scripts.
If you're comparing website proposals and want to see live projects before deciding, talk to Wama and ask for the portfolio URLs.
