Google Page Speed Analyzer
How to Use This Page Speed Checker
- Enter a URL — paste the full address, including
https://. - Run both mobile and desktop — they are separate assessments, and mobile is the one that counts for ranking.
- Read field data first, lab data second — field data is what Google measures; lab data is where you debug.
- Check all three metrics, not the score — a 90 on the Lighthouse score can still fail INP in the field.
- Fix the worst metric first — LCP is the hardest to pass and drags down the overall result more than the other two.
Field Data vs Lab Data: The Distinction Most Reports Get Wrong
PageSpeed Insights returns two entirely different measurements, and conflating them is the most common reason people optimize the wrong thing.
| Dimension | Field data (CrUX) | Lab data (Lighthouse) |
|---|---|---|
| Source | Real Chrome users who opted in | Simulated load on a single device |
| Time window | Trailing 28 days, 75th percentile | One snapshot, fixed network conditions |
| INP available? | Yes — this is the only place INP exists | No — TBT is the stand-in |
| What Google ranks on | This one | Not directly |
| Best used for | Tracking what users actually experience | Diagnosing specific causes |
The practical consequence: a lab test cannot produce an INP number because the simulated load never clicks, taps, or types. Lab tests report Total Blocking Time (TBT) instead, which is the best available proxy — but it covers only the load phase, while field INP captures every interaction during the whole visit, including menus, filters, and consent banners. TBT and INP correlate without mapping 1:1.
If field data is unavailable — common for low-traffic pages, which CrUX excludes to preserve anonymity — PageSpeed Insights falls back to lab data alone and says so explicitly. A page with no field data has no confirmed passing status, regardless of its lab score.
The 2026 Core Web Vitals Thresholds
The thresholds have held steady into 2026. A URL passes only when 75% of its visits hit the "Good" band on all three metrics, measured at the 75th percentile over a rolling 28-day window.
| Metric | What it measures | Good | Needs improvement | Poor |
|---|---|---|---|---|
| LCP — Largest Contentful Paint | How fast the main content appears | ≤ 2.5 s | 2.5–4.0 s | > 4.0 s |
| INP — Interaction to Next Paint | Latency of user interactions | ≤ 200 ms | 200–500 ms | > 500 ms |
| CLS — Cumulative Layout Shift | Unexpected movement after rendering | ≤ 0.1 | 0.1–0.25 | > 0.25 |
INP replaced First Input Delay (FID) on 12 March 2024. FID measured only the input delay before an event handler started; INP measures the entire interaction, from input to the next frame painted. That is why sites that passed comfortably under FID can fail under INP without changing a line of code.
2026 Pass Rates: Where You Actually Stand
Just over half the web passes all three metrics on mobile. The gap between mobile and desktop is large, and it widened year over year — from 14.2 percentage points to 16.5.
| Metric | Mobile pass rate | Desktop pass rate | Device gap |
|---|---|---|---|
| LCP | 63.7% | 78.4% | 14.7 pts |
| INP | 71.2% | 89.1% | 17.9 pts |
| CLS | 78.9% | 82.3% | 3.4 pts |
| All three | 51.3% | 67.8% | 16.5 pts |
Source: CrUX data, June 2026. Two things stand out. LCP is the hardest metric to pass at 63.7% on mobile, and it drags down the composite result more than INP or CLS because its failures cluster around large hero images, render-blocking resources, and slow server response — infrastructure problems, not code tweaks. CLS is the easiest and the closest to parity across devices, because layout shift is a CSS problem that affects every viewport equally.
Ecommerce is the worst-performing vertical: only 36% of the largest stores pass on mobile, versus about 65% of the smallest. Scale works against performance — more templates, more third-party tags, and more feature surface area.
Are Core Web Vitals a Ranking Factor?
Yes — Google has confirmed that page experience signals, including Core Web Vitals, are a ranking input, and field data at the 75th percentile from real Chrome users is the measurement source. But the nuance matters more than the headline:
- CWV works as a tie-breaker. When two pages address the same query equally well, the faster one is more likely to rank higher. It does not override weak content.
- Improvements help gradually — until they don't. Moving a metric from "Poor" toward "Good" produces a gradual benefit. Once all three are in the Good band, further optimization does not improve rankings.
- Google's March 2026 core update changed how the signals are combined. CWV is now evaluated as a composite score rather than three independent signals, which means a single badly failing metric weighs more heavily than it used to.
For AI search, the picture is different. Core Web Vitals are not a demonstrated citation-ranking factor across AI engines, even though they remain a real Google ranking input. Optimizing for AI answers requires entity consistency and extractable structure, not faster paint times.
How to Fix Each Metric
The fixes are metric-specific. Applying LCP tactics to an INP problem wastes effort, so diagnose the failing metric before choosing a technique.
LCP — The Hero Element Is Almost Always the Cause
- Find the LCP element in the lab report; it is usually the hero image, a headline, or a background video poster.
- Prioritize it with
fetchpriority="high"and avoid lazy-loading the LCP image — lazy-loading the hero is one of the most common self-inflicted LCP regressions. - Serve it in a next-generation format at the exact rendered dimensions, through a CDN.
- Reduce server response time — TTFB above 800 ms caps how fast LCP can be, because the browser cannot paint what it has not received.
- Eliminate render-blocking CSS and JS in the critical path, and consider server-side rendering for client-heavy frameworks.
INP — Long Tasks Block Every Interaction
- Keep tasks under 50 ms. Long tasks are the most common cause of poor INP because they block the main thread while the browser is trying to respond to input.
- Yield with
scheduler.yield()— available in Chromium since 2024, it pauses the current task and re-queues the remainder with high priority, giving the browser a frame to handle input. UsesetTimeout(fn, 0)as the universal fallback. - Move heavy work to Web Workers — parsing, sorting, and data transformation do not need the main thread.
- Debounce input handlers so keystrokes and filters do not queue a repaint per event.
- Watch the lab TBT number — keeping TBT under 200 ms is the practical Lighthouse target, though it is a proxy, not a guarantee of field INP.
CLS — Reserve Space Before Content Arrives
- Set explicit
widthandheighton every image and video, or use CSSaspect-ratio. - Reserve space for ads and embeds — dynamic slots that expand after load are the classic CLS cause.
- Use
font-display: swapand preload the primary font so late font swaps do not shift text. - Avoid injecting banners above existing content after first paint.
After fixing the failing metric, verify the page is actually indexed — a perfect Core Web Vitals profile on an unindexed page accomplishes nothing. Run it through our Google Index Checker before assuming the speed work will pay off.
The Business Impact: What the A/B Tests Show
Core Web Vitals work is measurable in revenue, not just in scores. Rakuten 24 ran a controlled A/B test where the only difference between the two versions was Core Web Vitals optimization — no functional or visual changes:
| Metric | Change |
|---|---|
| Revenue per visitor | +53.37% |
| Conversion rate | +33.13% |
| Average order value | +15.20% |
| Exit rate | −35.1% |
The A/B methodology is what makes this credible: nothing changed except the vitals. At the other end of the funnel, the abandonment curve is steep — going from a 1-second to a 10-second load raises the probability of a mobile bounce by 123%, and a single second of delay reduces conversions by up to 20%, with bounce rates climbing 32% between the 1- and 3-second marks.
If your site monetizes through ads, slower pages also reduce ad viewability, which reduces RPM. Estimate that effect with our AdSense Revenue Calculator after the speed fixes are deployed.
Frequently Asked Questions
Is this page speed checker free?
Yes. The tool is completely free, requires no sign-up, and runs in your browser. The URLs you test are not stored on a server.
Why does INP show "N/A" in the lab results?
Because a simulated load never clicks, taps, or types. INP is an interaction metric, so it can only be collected from real users. Lab tests report Total Blocking Time (TBT) instead, which covers the load phase only and does not map 1:1 to field INP.
What are the 2026 Core Web Vitals thresholds?
LCP ≤ 2.5 s, INP ≤ 200 ms, and CLS ≤ 0.1 for a "Good" rating. A URL passes when 75% of visits hit the Good band on all three, measured at the 75th percentile over a rolling 28-day window.
Does page speed actually affect rankings?
Yes — Core Web Vitals are a confirmed ranking input and act as a tie-breaker between pages of similar content quality. But once all three metrics are in the Good band, further optimization does not improve rankings. Content relevance still dominates.
Which metric should I fix first?
LCP. It has the lowest mobile pass rate (63.7%) and drags down the composite result more than INP or CLS. Google's March 2026 update also made CWV a composite score, so a single badly failing metric now weighs more heavily than before.
What is a good PageSpeed Insights score?
The Lighthouse score is a lab diagnostic, not a ranking signal. A 90 can still correspond to a failing field INP, and a 70 can correspond to a passing field assessment if the slow lab conditions do not reflect real users. Track the three field metrics, not the score.
Related Performance & Technical SEO Tools
Use these free tools to fix the issues that speed testing exposes:
- Google Index Checker — confirm the page you optimized is actually indexed.
- Broken Link Checker — dead links waste crawl budget that could reach your faster pages.
- AdSense Revenue Calculator — estimate the RPM impact of improved ad viewability.
- SEO Site Score Generator — see whether speed is one of several compounding signal gaps.
- Meta Tag Analyzer — fast pages still need titles that survive Google's 76% rewrite rate.