How do I check if my website is actually fast?

Check it on a mid range phone over mobile data, first visit, and count. Then run the same URL through PageSpeed Insights. A dental clinic menu and a restaurant page both fail or pass on that clock, not on a laptop in the office.

I will not read your score from here. I will tell you how I timed a clinic set, and how you can time yours.

The phone test, the one that matches the leave

Turn off wifi. Use 4G or 5G you actually get on the road. Close the tab. Open the URL as if you had never been there.

Count. People leave after three seconds. The home page already uses that line. I will not invent a kinder clock for a shop I like.

A Kalkaji restaurant owner who timed the menu on a Redmi, standing outside the shop on the same 4G the diner has, learned the slider was the wait. The office Mac had hidden it. A Defence Colony clinic that only ever opened the site on clinic wifi thought the page was fine. On the road it was still a spinner past three.

If you are still on a spinner, the site is slow for the stranger. If the page painted and you still cannot see hours or a fee, the wait was not the only hole. Why people leave in three seconds splits those two. Why the website is slow on mobile is the file pile. Ads or fix the website first is the spend that points at a spinner.

Google’s writing on why speed matters says speed affects whether people stay and whether they convert. It does not give your restaurant a number. It is the reason the count is worth doing.

The lab test, the one that gives a score

PageSpeed Insights is Google’s public tool. Paste the URL. Read the mobile run. I use that family of scores on the clinic set I already publish: middle 8.8s, 2.4 MB, PageSpeed 54. The sample website I built: 1.7s, 89 KB, Lighthouse 100.

Those two bars are evidence I have. They are not a Delhi average. They are not your gym. If your result sits in the eight to ten second band, you are in the shape I keep meeting. If you are already near the sample, speed is probably not the rebuild.

I will not tell you a single Lighthouse number buys a rank. A faster score does not purchase a Maps pack. Google will not sell you a better local slot. Speed is a stay-or-leave problem first.

Write the mobile score next to the seconds you counted. If the tool is green and the phone is still spinning, trust the phone. The lab is a lab. The diner is on the road.

What to write down, so the test is not a mood

• Seconds you counted on the phone.
• Weight and score the tool showed, if it showed them.
• Whether hours and a next step were visible when the page finally painted.

Bring those three to anyone you might pay. A person who only sends a laptop screenshot is not testing the stranger.

Do the count twice if the first run felt lucky. First visit means the files were not already sitting on the phone. A second open from the same tab is a different, kinder test. Use the first open when you argue with a quote.

What I will not tell you

I will not tell you to buy a new theme to chase 100. A new coat on the same files often stays slow.

NexBuilt will not tell you 1.7s is a promise made to every URL. It is a sample I measured. Floor for a NexBuilt site is Rs 30,000 on pricing. Usually under two weeks. Those are the published bars, not a score.

I will not tell you a restaurant in Kalkaji and a clinic in Defence Colony share one magic number. Weigh the URL you actually have. Paste that URL, not a demo.

If you already run ads at that URL, do the phone count before the next rupee. A spinner that you paid to show is still a spinner. The three lines you write tonight are enough to pause the spend. A prettier theme next month will not refund the clicks that already left.

What to do with the three lines you wrote

If the count failed, stop pointing ads at the URL until the wait is honest. If the count passed and the next step is missing, the job is the path, not the theme.

Send the three lines. I will say the wait is the rebuild, the path is a smaller fix, or both already pass and you can leave the URL alone.