Page speed is the grade PageSpeed Insights assigns to one URL. Google runs the check as a lab test on a simulated mid-tier phone and a mobile network, and the report shows 28 days of real Chrome visits when that page has enough of them.
A plumbing company in East Atlanta can see a score under 50, which Google marks as poor, because the van photo at the top is enormous. The call link can already be on the screen. The grade and the tap answer different questions.
What Page Speed Is Measuring
The number at the top of the lab section is a Lighthouse performance score. A score of 90 or above is good, 50 to 89 needs improvement, and anything under 50 is poor. Real visits stay a separate question even when that lab score is green, which the same guide states outright.
When the real-user section is present, it comes from the Chrome User Experience Report. It covers the previous 28 days and reports the 75th percentile, so a quarter of visits were that slow or slower.
When a page is new or quiet, the same guide says the report may fall back to the whole site, and the real-user section stays empty only if the site is short on samples too. The lab is the rehearsal you can rerun after one change.
The Wait in Front of the Money Action
Three waits show up on a typical small-business homepage. Change the one that sits in front of Call, Book, or the form.
- Hero file: The largest thing on the first screen, usually a photo exported straight from a phone. Largest Contentful Paint often waits on that file, so the picture has to be small before it is uploaded. Lazy loading holds back pictures farther down the page.
- Third-party script: A chat bubble, a pixel, a booking embed, or a font hosted on someone else’s domain. In the lab, Total Blocking Time is the total time the page is too busy to respond. If Book sits behind that script, the customer feels the wait, and if Call is plain HTML with the script lower down, the score can look harsher than the visit.
- Layout jump: A banner or image with no height reserved. Cumulative Layout Shift is the score for that jump, and the thumb can hit the wrong target. Giving the image a set space keeps the button where it was.
A CDN shortens the trip for the file you already uploaded. The van photo still has to be smaller before that trip gets short.
Run the Same URL Twice
Open the homepage on your phone, on cellular, away from the office Wi-Fi. Count the seconds until Call, Book, or the form can be tapped. That count is closer to the visit your customer gets.
Then paste the same URL into PageSpeed Insights and stay on the mobile report. If real-user data is there, read the 75th percentile for Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift ahead of the big number. Those three are the Core Web Vitals, and Interaction to Next Paint shows up in that field section.
The lab’s busy-page signal is Total Blocking Time, and the score can shift between runs even when the page did not change. Change one thing: the photo, or one script you can live without for a week. Run the same URL again and keep the change that made the tap shorter.
Bring the URL and the Screenshot
Write down the cellular count and save the mobile screenshot. Note whether the real-user section was empty. That note tells the next person whether they are looking at a crowd or a rehearsal.
If the homepage is how jobs arrive, that wait is part of the product. We build that kind of site on our website design page.
Bring the URL and the screenshot. Fifteen minutes on the calendar is enough to name the wait to change first.
