Google PageSpeed Insights is Google's free tool that reports two separate things for a URL: what real visitors experienced over the previous 28 days, and a simulated Lighthouse test scored from 0 to 100. The real-visitor block is the one that matters for SEO in South Africa, and the big green or red number is the one most owners read first and trust too much.

This guide shows how to read both blocks in order, how the 0 to 100 score is built from weighted metrics, why it moves between runs, and what a South African site owner should do with the result. It leans on Google's own documentation throughout, and links to it where a claim depends on it.

Quick Answer

Read the real-user Core Web Vitals block at the top of Google PageSpeed Insights first: it covers the previous 28 days and decides whether the page passes. Treat the Lighthouse score underneath as a lab diagnostic, because it comes from one simulated run and is built from five weighted metrics, with Total Blocking Time carrying the most weight at 30%.

Red score and no idea which part to fix first?

Send us the URL and we will tell you which block of the report deserves your attention before you pay a developer.

Request a free audit

Google PageSpeed Insights Results: Field Data and Lab Data

A Google PageSpeed Insights report combines two data sources: field data from real users and lab data from a Lighthouse test. According to Google's PageSpeed Insights documentation, the two answer different questions, and Google describes the trade-off this way: "Lab data is useful for debugging issues, as it is collected in a controlled environment. However, it may not capture real-world bottlenecks. Field data is useful for capturing true, real-world user experience - but has a more limited set of metrics."

BlockWhere it comes fromWhat it tells youBest for
Field data (real-user experience)Real users' experiences over the previous 28-day collection periodWhether actual visitors get good loading, interaction and layout stabilityDeciding if you have a problem at all
Lab data (Lighthouse diagnostics)One simulated page load, with a mid-tier device on a mobile network for the mobile tab and an emulated desktop on a wired connection for the desktop tabA repeatable test and a 0 to 100 performance scoreFinding out what to fix and checking a fix

The PageSpeed Insights field data block reports First Contentful Paint (FCP), Interaction to Next Paint (INP), Largest Contentful Paint (LCP) and Cumulative Layout Shift (CLS), plus an experimental Time to First Byte. The lab block gives you the score and a list of audits. Two data sources on one screen is why people argue about whether a site is "fast": they are quoting different blocks.

Google also explains why the two disagree. According to web.dev's lab and field comparison, a lab test uses a single device on a single network from a single geographic location, while field data reflects the real devices, networks and locations of your users. Lab tests usually start with an empty cache. The same page adds that a lab cannot predict when users will choose to interact with a page; its wording on that point names the older First Input Delay metric, and it describes Total Blocking Time as a lab metric "intended to help diagnose issues with INP".

Which Number Should You Trust?

The number to trust is the field data when the report shows it, the origin-level field data when the page itself lacks enough traffic, and the lab score only as a diagnostic when neither exists. The table below maps what you see on the report to what it means and where to look next.

What the report showsPossible meaningHow to tell, and what to do
Field data for this URL, assessment passed, lab score orangeReal visitors are fine; the lab run is harsher than your typical visitor or hit one slow resourceCheck the Core Web Vitals at the 75th percentile. If LCP, INP and CLS are all good, the page passes (when INP data is thin, LCP and CLS alone decide it); use the lab audits for polish, not urgent repair
Field data for this URL, assessment failed, lab score greenA lab run cannot reproduce what real users hit: slower devices, interactions, shifts after load, or a cold cacheFind which metric is failing in the field block and chase that one. Web.dev notes lab CLS counts only above-the-fold shifts during load, while field CLS covers the page's whole lifespan
Field data only for the whole originThe page is new or has too few real-user samples, so PSI has fallen back to origin-level dataRead it as the average experience across your site's pages, not this URL. Fix the weakest template (product, category, blog) rather than one page
No field data at all, lab score onlyLikely too little real-user traffic for either level to be reported (an inference from the fallback rule, not a stated cause)Use the lab audits as a to-do list and re-test after changes; there is no real-user verdict to compare against yet

Google's documentation confirms the fallback: "A page might not have sufficient data if it has been recently published or has too few samples from real users. When this happens, PSI will fall back to origin-level granularity, which encompasses all user experiences on all pages of the website." A small South African business site with modest traffic may well land on origin-level data, and that is not a fault in your site.

Takeaway: field first, lab second

If the report shows real-user data, the pass or fail verdict comes from that block. The lab score is a diagnostic from a single simulated run, useful for deciding what to fix and for checking whether the fix worked.

How the 0 to 100 Performance Score Is Built

The Google PageSpeed Insights score is a weighted average of five lab metrics, each converted to a 0 to 100 score and then combined. Per Chrome's Lighthouse scoring documentation, the weights in its Lighthouse 10 table, the version that page leads with, are:

Lab metricWeight in scoreWhat it reflects
Total Blocking Time (TBT)30%How long the page is too busy to respond during loading
Largest Contentful Paint (LCP)25%When the biggest visible element finishes rendering
Cumulative Layout Shift (CLS)25%How much the layout jumps around
First Contentful Paint (FCP)10%When anything first appears
Speed Index10%How quickly the visible page fills in

Two facts follow from that table. First, only metrics feed the score: in Google's words, "only metrics contribute to your Lighthouse Performance score, not the results of Opportunities or Diagnostics." Clearing a long list of suggestions does not raise the number unless it moves one of the five metrics. Second, INP is not in the list. INP is a Core Web Vital in the field block, but the lab score has no INP component, so a page can score green in the lab and still fail INP with real users. web.dev makes the same point: "A page may have poor INP despite having very good TBT scores."

The weights as arithmetic

The five weights sum to 100%: 10 + 10 + 25 + 30 + 25 = 100. TBT, LCP and CLS together carry 30% + 25% + 25% = 80% of the score; FCP and Speed Index together carry the remaining 20%.

That means one metric can cap the score. If LCP, CLS, FCP and Speed Index all scored a perfect 100 and TBT scored 0, the total would be 25 + 25 + 10 + 10 + 0 = 70 points, an orange score. If TBT scored 50 instead and everything else scored 100, you would lose 30% of 50 = 15 points and land at 85. Each metric is first converted to its own 0 to 100 score; Lighthouse's calculator lets you try values yourself.

The score bands are fixed: 90 to 100 is green and good, 50 to 89 is orange and needs improvement, and 0 to 49 is red and poor. Each metric value is mapped to 0 to 100 on a fixed curve whose control points come from HTTP Archive data: per Chrome's documentation, the 25th percentile of that data becomes a score of 50 and the 8th percentile becomes a score of 90. As a judgement call, chasing 100 is rarely the best use of time once the real-user block passes.

Reading the Core Web Vitals Assessment

The Core Web Vitals assessment is a pass or fail verdict that depends on three field metrics, not on the Lighthouse score. In Google's documentation, "the aggregation passes the Core Web Vitals assessment if the 75th percentiles of all three metrics are Good." There is one exception that matters for smaller sites: "If the aggregation has insufficient data for INP, then it will pass the assessment if both the 75th percentiles of LCP and CLS are Good." The verdict applies to whichever aggregation the report shows, the URL or, after a fallback, the whole origin, over the previous 28 days.

Core Web VitalGood thresholdWhat a miss usually points to
Largest Contentful Paint (LCP)Within 2.5 seconds of the page starting to loadHeavy hero image, slow server response, render-blocking resources
Interaction to Next Paint (INP)200 milliseconds or lessHeavy scripts, third-party apps and tags competing for the browser
Cumulative Layout Shift (CLS)0.1 or lessImages without set dimensions, late-loading banners and embeds

These PageSpeed Insights Core Web Vitals thresholds come from web.dev's Core Web Vitals page, which recommends measuring at the 75th percentile of page loads, segmented across mobile and desktop. The third column is a practitioner's shortlist of likely causes, not a diagnosis: confirm with the lab audits or Chrome DevTools before spending money.

The 75th percentile matters because a pass means three in four visits met the threshold, not that every visit did. A handful of slow connections will not fail you. A consistent pattern of them will. Your Core Web Vitals in South Africa status can also differ between mobile and desktop, because the metrics are segmented by device. As a judgement call, open the PageSpeed Insights mobile score tab first, since it simulates a mid-tier device on a mobile network. Check your own analytics for the device split before you decide how much the desktop tab matters.

Why the Score Moves Between Runs

A Lighthouse score moves between runs because it is a measurement taken in changing conditions, not a fixed property of your site. Chrome's documentation says performance fluctuations usually come from changes in underlying conditions, and lists A/B tests or changes in the ads being served, internet traffic routing changes, testing on different devices, browser extensions and antivirus software. It recommends thinking of site performance "as a distribution of scores, rather than a single number."

For a South African site there are two further points to hold in mind. Neither is a rule, and both are hypotheses worth checking before you act on a surprising score.

ObservationHypothesesHow to separate them
Score drops or jumps between two tests of the same URLAd or A/B-test variation, network routing changes, a heavy third-party script behaving differently, a different test locationRun the test several times and look at the spread. Compare the lab number with the field block, which does not vary with each test run (it is a rolling 28-day window)
Server response looks slow in the lab but visitors seem finePSI reports running from North America, Europe or Asia, so distance to your host may be part of the lab result; or a cache is warm for real users and cold in the labCheck your field data. Ask your host where the server is and whether a CDN or page cache sits in front of it
Mobile score far below desktopThe mobile tab simulates a mid-tier device on a mobile network; desktop uses an emulated desktop on a wired connectionCheck the Core Web Vitals report in Search Console for mobile and desktop separately before treating the gap as a defect

Google's PageSpeed documentation states that PSI "will report running in one of: North America, Europe, or Asia", and South Africa is not among them. The mobile lab run also uses a simulated mid-tier device on a mobile network. Together, that means the lab result may overstate or understate what visitors in Johannesburg, Durban or Cape Town actually see, and only field data can settle which.

The practical South African decision is about hosting. If your server sits in South Africa and your visitors are local, a lab run from overseas adds distance that they do not experience, so a slow server-response audit deserves a second look against your field LCP. If your host is overseas, that distance is real for your visitors too, and a CDN or page cache in front of the site is the first thing to ask your host about. That reading is a hypothesis to test, not a documented rule.

What to Do After You Read the Report

The right next step is to fix whichever field metric is failing, using the lab audits to find the cause, then re-test and wait for the 28-day window to roll forward. Field data is a rolling 28-day collection, so a fix you ship today will not fully show in the real-user block for weeks.

A workable order for a small business site:

  • Identify the failing Core Web Vital in the field block, on mobile first. If nothing fails, stop tuning and put the time into content or conversion.
  • Use the Search Console report for scale. It is based on real world usage data and groups URLs with a similar experience, so it shows whether a problem is one page or a whole template. Only indexed URLs appear, and groups without enough data are omitted.
  • Match the metric to the likely cause with the lab audits, then fix the largest cause first: image weight and server response for LCP, third-party scripts for INP, set dimensions for CLS.
  • Re-run the lab test several times after the change to confirm direction, and recheck the field block after the 28-day window has moved on.

Our page speed SEO checklist and the broader guide to website speed optimisation cover the fixes themselves. If your store runs on Shopify, Shopify speed optimisation deals with theme and app weight, and page speed and conversion covers why the commercial case sits behind the technical one.

Do Core Web Vitals Affect Rankings?

Good Core Web Vitals are something Google says it wants to reward, but a passing result does not buy a position. Google's Search Central page says Core Web Vitals "aligns with what our core ranking systems seek to reward" and recommends that site owners "achieve good Core Web Vitals for success with Search". It does not say a passing page outranks a competitor with better content, so treat speed as one input among many. The same logic is why a page that passes should not absorb more of your budget than a weak title, thin content or a missing page.

Takeaway: the report in four reads

Read the field verdict first, then identify the failing metric, then use the lab audits to find the cause, and last check the score. A change counts as a success when the real-user block moves, not when one lab run turns green.

If you are building a wider checklist, the DIY SEO audit places speed in context, and the free SEO tools list covers the other free checkers worth keeping next to this one. For how pages are crawled and ranked in the first place, see how Google Search works.

Want the speed fixes ranked by what they are worth?

Tell us your site and your slowest page, and we will tell you what to look at first. No obligation.

Book a free audit

Why South African Businesses Choose Growth Pulse Media

A speed report is only worth acting on when someone ties it to what the site earns, and that is how Growth Pulse Media, a Johannesburg agency, approaches its SEO services, starting from Search Console data rather than a single lab score. Founder Dirk van Greuning built and scaled a large South African ecommerce business before the agency, so the question we ask about speed is a commercial one: does the fix move something that earns money? All work is executed in-house, with a limited client load for senior attention.

Who This Is NOT For

This guide is not for owners hoping a single score will settle every speed argument, sites with almost no traffic, or anyone expecting speed work to rescue weak content.

Owners chasing a perfect 100. The score is a weighted lab average, and the field block decides the Core Web Vitals verdict. Once the real-user assessment passes, extra lab points are, as a judgement call, rarely the best use of a developer's hours.

Sites with very little traffic. A new or low-traffic page may show only origin-level data or none at all, so there is no real-user verdict to read. Use the lab audits as a to-do list and put your effort into getting visitors first.

Anyone treating speed as a ranking shortcut. Google describes Core Web Vitals as aligned with what its ranking systems reward, not as a guarantee of position. Weak content stays weak on a fast page.

Frequently Asked Questions

What is a good Google PageSpeed Insights score?

Google treats 90 to 100 as good, 50 to 89 as needing improvement and 0 to 49 as poor. That is the lab score, though, and the pass or fail Core Web Vitals verdict comes from real-user field data. A page can pass the assessment with an orange lab score.

Why is my mobile score lower than my desktop score?

The mobile test simulates a mid-tier device on a mobile network, while the desktop test uses an emulated desktop on a wired connection. A lower mobile score is expected, so check the mobile field data and the Search Console report before treating the gap as a fault.

What is the difference between field data and lab data?

Field data is what real users experienced over the previous 28-day period, and lab data is a single simulated test in a controlled environment. Field data is the better guide to real-world experience, while lab data is better for debugging and checking a fix.

Why does my score change between tests?

Lighthouse scores shift with changing conditions: A/B tests or ads, network routing, the device used, browser extensions and antivirus software. Google suggests viewing performance as a distribution of scores rather than a single number, so test several times and compare against field data.

Does PageSpeed Insights show data for South Africa?

The field block reflects real Chrome users of your page over the previous 28 days, wherever they are. The lab test runs from North America, Europe or Asia according to Google's documentation, so its location is not South African. If your page has too little data, PSI falls back to origin-level data or shows none.

Which metrics count towards the Lighthouse score?

Five lab metrics, using the weights in Lighthouse 10: Total Blocking Time at 30%, Largest Contentful Paint at 25%, Cumulative Layout Shift at 25%, First Contentful Paint at 10% and Speed Index at 10%. Opportunities and Diagnostics do not feed the score, and INP is not one of the five.

Get Your Speed Report Read by a Person

We start from Search Console, read the field data first and tell you what is worth fixing. No obligation — we'll get back to you within 24 hours.

Request Your Free Audit
Dirk van Greuning — Founder, Growth Pulse Media
Dirk van Greuning Founder, Growth Pulse Media

Founder of Growth Pulse Media and a specialist in South African search dominance. Dirk translates his experience in scaling South African businesses into high-velocity digital strategies for B2B and retail leaders. He writes about SEO, lead generation, and paid media from an operator's perspective — prioritising pipeline value over impressions.

Connect on LinkedIn