
PageSpeed Insights (PSI) is Google's free report on how fast a web page loads for real visitors and in a lab test. The number at the top matters less than the three timing measures underneath it, and most small business sites can improve those with a handful of specific fixes. A site that is built lean from the start, and kept that way as content changes, avoids most of them. That is how we build every site at Cincinnati Web Designs: by hand, without themes or page builders.
This guide shows you how to run the test, what each part of the report means, which scores are good, and what to fix first.
TL;DR
- PSI shows two things: field data from real Chrome users over the past 28 days, and a lab test run once, right now.
- Check the three Core Web Vitals first: LCP (loading) within 2.5 seconds, INP (responsiveness) under 200 milliseconds, CLS (layout shift) under 0.1.
- Lighthouse scores of 90 or above are rated good, 50 to 89 need improvement, and below 50 are poor.
- Speed is one part of Google's page experience signals. It will not outrank relevant content, but a slow page makes visitors leave and gives you a weaker position when many helpful results compete.
- The biggest wins are usually the hero image, third-party scripts and bloated page code.
Table of Contents
- What PageSpeed Insights actually measures
- How to run a PageSpeed Insights test
- Reading the score: what is a good PSI result?
- Does page speed affect SEO?
- The fixes that move the numbers
- Other speed test tools, and when to use them
- What the report does not tell you
- A simple monthly routine
- Our perspective
- Get a fast site without managing it yourself
- FAQ
- Sources
What PageSpeed Insights actually measures
PSI is Google's performance tool for both mobile and desktop. It combines two kinds of data, as Google's PSI documentation explains:
- Field data comes from the Chrome User Experience Report (CrUX), which collects measurements from real Chrome users who visited your page. It covers a rolling 28-day window.
- Lab data comes from Lighthouse, a testing engine that loads your page once under controlled conditions and produces diagnostics.
The difference matters. Field data tells you what customers experienced. Lab data tells you why, and lets you check a fix within minutes. Google notes that lab results need not match real-user performance, so a perfect lab score does not promise that every visitor has a fast experience.
Field data is also not real time. Google's PSI FAQ says it reflects a trailing period and is updated daily, while the lab analysis can reflect your changes within minutes. If you fix something today, expect the lab score to move today and the field data to catch up over the following weeks.
How to run a PageSpeed Insights test
The steps are short, but each one has a spot where people misread the result.
- Open PageSpeed Insights and paste a full page address. Start with your home page, then test your most important service page. Do not stop at the home page: each URL gets its own result.
- Press Analyze and wait. The tool loads the page and builds the report.
- Start on the Mobile tab. Most local searches happen on phones, and the mobile test is the stricter one. Then check Desktop.
- Look for the field data section at the top. If it says a Core Web Vitals assessment passed or failed, that is the real-visitor verdict.
- Scroll to the lab results and diagnostics. This is where the specific fixes are listed.
- Repeat after each change. Run the test two or three times, because a single lab run varies a little with network conditions.
Page or whole site: which result are you reading?
Google's FAQ distinguishes a single-URL result from an origin-level result, which combines speed data across the pages served from your domain. A strong origin result can hide a slow page, and a slow page can sit inside an otherwise healthy site. Check both, and treat the slow individual page as the one to fix.
What if there is no field data?
CrUX data may be unavailable for a URL, usually because too few Chrome visitors reached it. Google's guidance in that case is to rely on the Lighthouse synthetic audit to estimate performance and review its recommendations. A new service page or a small local site will often land here. It does not mean something is wrong.
Reading the score: what is a good PSI result?
Google rates Lighthouse category scores of 90 or above as good, 50 to 89 as needing improvement, and below 50 as poor, per the PSI documentation. That number is a lab score. The Core Web Vitals in the field data are what Google's page experience guidance refers to.
| Metric | What it tracks | Good | Needs improvement | Poor |
|---|---|---|---|---|
| LCP (Largest Contentful Paint) | How long until the biggest element, often a hero image or headline, appears | 2.5 seconds or less | 2.5 to 4 seconds | Over 4 seconds |
| INP (Interaction to Next Paint) | How quickly the page responds when someone taps or clicks | Under 200 ms | 200 to 500 ms | Over 500 ms |
| CLS (Cumulative Layout Shift) | How much content jumps around while loading | Under 0.1 | 0.1 to 0.25 | Over 0.25 |
The good thresholds for all three come from Google's Core Web Vitals guidance, and the LCP bands from the PSI documentation. INP replaced the older First Input Delay metric when Google announced it on May 10, 2023. If you want a deeper walkthrough of the loading metric specifically, there is a step-by-step guide on how to improve LCP.
The 75th percentile rule
PSI reports the 75th-percentile value for each metric, plus the share of visits that were good, needing improvement and poor. Google's Web Vitals measurement guide says 75% of page visits should meet the good threshold for each metric for a page to pass. In plain terms: your page passes when three out of four visitors get a good experience on all three measures.
A worked example
Say a plumber's home page shows a mobile lab score of 62, an LCP of 3.4 seconds, and a CLS of 0.02. Interaction is fine and layout is stable. The problem is almost entirely loading. The diagnostics point to a 2 MB hero photo. Resize and compress that one image, and the LCP drops toward the good range, while the score rises with it. You fixed one thing, not forty.
That is the right way to use the report: find the one or two audits that account for the slow metric, and ignore the rest until those are done.

Does page speed affect SEO?
Yes, but with the right expectations. Google states that Core Web Vitals are used by its ranking systems, yet they are one part of the broader page experience, and good scores do not guarantee top rankings. Its core ranking systems prioritize relevant content even when page experience is subpar. A good page experience can contribute to success when many helpful results exist.
For a Cincinnati contractor, dentist or restaurant, that last condition describes the local results page well: many businesses with similar content competing for the same searches. Speed will not rescue a thin page, and a thin page will not be rescued by a perfect score. But when your content matches the competition, a faster page is one of the few advantages you fully control. For the business side of this, see our piece on how slow sites lose Cincinnati customers, and why a great website alone isn't enough to win on Google.
There is also a plain customer reason. A visitor on a phone with a leaky roof or a dead water heater will not wait for a heavy page. Speed is a conversion issue before it is a ranking issue.
The fixes that move the numbers
Work in this order, because the early items usually carry the most weight.
1. The hero image
The largest element on most local business pages is a big photo at the top. Serve it at the size it is displayed, in a modern compressed format, and make sure the browser knows it is the priority. Google's Fetch Priority guidance describes a Google Flights test where marking the main image as high priority improved LCP from 2.6 seconds to 1.9 seconds. Do not lazy-load that top image: lazy loading belongs to images further down the page.
2. Third-party scripts and tags
Chat widgets, analytics, ad pixels, heat maps and review badges each add weight. Google's tag guidance explains that tags can use bandwidth and main-thread time, potentially worsening LCP, CLS and INP, and recommends periodically auditing which tags you actually use. Remove anything nobody has looked at in six months.
3. Layout shift
A button that jumps just as someone goes to tap it is a CLS problem. Give images and embedded maps fixed dimensions so the browser reserves space, and avoid inserting banners above content after the page has started to load.
4. Code weight
Themes and page builders often ship large amounts of CSS and JavaScript for features a page never uses. PSI flags this as unused code. The cleanest fix is a page that only contains the code it needs. We write that way: every site we build is hand-coded rather than assembled from themes or page builders.
5. Hosting response time
If the server is slow to answer, every other metric starts late. The PSI field data includes an experimental time-to-first-byte measure, which is a clue here. Quality hosting is part of the website plan we offer, not a separate purchase.
Other speed test tools, and when to use them
PSI is the right first stop, but it is not the only view. Google's measurement guide points to two more:
- Chrome DevTools, Performance panel: for debugging a specific slow moment on a page, built into the Chrome browser.
- Search Console's Core Web Vitals report: monitors your verified site by page group and keeps historical data, so you can see whether a fix held.
Third-party speed testers exist as well and are useful for waterfall charts of what loads in what order. Their scores will not match PSI exactly, because each uses different test locations and settings. Pick one testing method, record your baseline, and compare like with like. Switching tools between runs makes it look as if you improved or slipped when you did neither.
What the report does not tell you
- It will not rank your pages. A fast page with nothing useful on it stays invisible. Content, service pages and local signals come first. See our guides on service page SEO and the on-page checks local businesses should fix first.
- A perfect 100 is not the goal. Passing Core Web Vitals in the field data and having a fast, usable page is the goal. Chasing the last few points can mean removing features your customers use, such as a booking form or a map.
- It is a snapshot. Pages slow down again when someone adds a new plugin, a bigger image or another tracking tag. Speed is maintenance, not a one-time project.
A simple monthly routine
- Test your home page and top three service pages on mobile.
- Note LCP, INP and CLS in a spreadsheet.
- If any slipped into needs improvement, look at what changed on that page in the last month.
- Fix the largest item in the diagnostics, then retest.
- Check Search Console's Core Web Vitals report for trends.
Ten minutes a month catches most regressions before customers notice. If you also track calls, form fills and map views, local SEO reporting shows how to put speed next to the numbers that pay the bills.
Our perspective
We build websites for Cincinnati businesses, and speed problems almost always trace back to the same cause: a site assembled from heavy themes and add-ons, then loaded with extra scripts over the years.
That is why we hand-code every site rather than start from a theme or page builder. A page that only carries the code it needs has less to fix later. Hosting, SSL, your domain and ongoing maintenance are included in the $99 a month website plan, and so are unlimited content updates, so you are not paying extra each time you change a photo or a phone number.
Use PageSpeed Insights as the dashboard it is. Read the three Core Web Vitals, fix the largest cause, and retest. Your goal is a site that is quick for the person on a phone in Over-the-Rhine or Mason who needs you today.
Yoshua
Get a fast site without managing it yourself
If your report shows slow mobile loading and you would rather hand the work to someone, our Website plan is $99 per month with no upfront development cost, and it covers a hand-coded, mobile-responsive site with hosting, SSL, your domain, maintenance and unlimited content updates. Most of our offerings run on a 12-month contract. Already have a site that scores poorly? Our website redesign is also $99 per month. See everything on the pricing page.

Want to see what a lean build looks like in practice? Read the Cin City Moto case study.
FAQ
Is PageSpeed Insights accurate?
It is accurate for what it measures, but the two halves measure different things. Field data reflects real Chrome visitors over a trailing period, while the lab score is one test under set conditions. Use field data to judge the real experience and the lab results to find and verify fixes.
What is a good PageSpeed Insights score?
Google rates a Lighthouse score of 90 or above as good, 50 to 89 as needing improvement, and below 50 as poor. For Core Web Vitals, aim for LCP within 2.5 seconds, INP under 200 milliseconds and CLS under 0.1 for at least 75% of visits.
Why is my mobile score lower than my desktop score?
The mobile test simulates a slower device and connection, so large images, heavy scripts and long code take more time to process. Since most local searches happen on phones, treat the mobile result as the primary one and fix it first.
Why does my score change every time I test?
Lab tests run under slightly different network and server conditions each time, so small swings are normal. Run the test two or three times and look at the trend. Field data changes slowly because it covers a rolling window of real visits.
Does page speed really affect my Google rankings?
Core Web Vitals are part of Google's ranking systems, but Google says relevant content still comes first and good scores do not guarantee top positions. Speed helps most when many helpful pages compete for the same search, which is common in local results.
Sources
- About PageSpeed Insights, Google for Developers
- PageSpeed Insights FAQ, Google for Developers
- Understanding Core Web Vitals and Google search results
- Understanding Google Page Experience
- Getting started with measuring Web Vitals, web.dev
- Optimize resource loading with Fetch Priority, web.dev
- Best practices for tags and tag managers, web.dev
- How to improve LCP, step-by-step guide