The mark Google uses
Google publishes three measures it calls Core Web Vitals. The one a visitor feels first is how long the main content takes to show. In Google's words: "A good LCP value is 2.5 seconds or less." The second is how quickly the page answers a tap: "pages should have a INP of 200 milliseconds or less".
These are judged on real visits, not on your best day: "a good threshold to measure is the 75th percentile of page loads". Three visitors in four should get a page that fast.
What we measured
In November 2026 we timed 191 Twin Cities roofing homepages on a phone-size screen with a slowed processor and a slow mobile connection. 158 took longer than 2.5 seconds to show their main content. The middle one took 7.4 seconds. It is one trade and a lab test, and it is the only data on this page that is ours.
What the slow pages had in common
They were heavier. This table shows the middle value in each group for the whole page, for its scripts, and for the number of files requested.
| Main content showed in | Pages | Page weight | Scripts | Files |
|---|---|---|---|---|
| 2.5 seconds or less | 33 | 541 KB | 59 KB | 49 |
| 2.5 to 10 seconds | 86 | 1,210 KB | 219 KB | 62 |
| Over 10 seconds | 72 | 2,265 KB | 382 KB | 84 |
Weight is not the only cause of a slow page, and these are weights as the browser reported them. But a page four times heavier has four times as much to send down a phone connection, so it is the first place we look.
What the main content was
The browser reports which piece of the page it counted as the main content. For 111 of the 191 pages that was clearly a video, a picture or text.
| Main content was | Pages | Middle time | Within 2.5 seconds |
|---|---|---|---|
| A video | 16 | 11.9 seconds | 0 |
| A picture | 48 | 5.1 seconds | 6 |
| Text | 47 | 2.7 seconds | 22 |
Not one of the 16 pages that led with a video showed it within 2.5 seconds. A background video is the most expensive thing you can put at the top of a page.
A lab test is not a customer
Our numbers come from a lab: one simulated phone, one fixed connection. Google is plain about the limits of that: "Lab data is useful for debugging issues, as it is collected in a controlled environment. However, it may not capture real-world bottlenecks." Your real visitors use many phones and many connections. A lab number tells you what to fix. It does not tell you what every visitor sees.
How to test your own site
Open PageSpeed Insights, type your address and choose Mobile. Read the line called Largest Contentful Paint. If Google has enough real visits to your site, it shows those first. For a small site it often does not, and you get lab numbers only.
Then do the plain test: stand outside, turn off wifi, open your homepage and count the seconds until you can read it.
What to fix first
Start at the top of the page, because that is what the visitor waits for. A video behind the headline: replace it with one picture. A very large picture: resize it for a phone. Scripts for chat boxes, pop-ups and trackers you no longer use: remove them. None of this needs a new website, though an old one built from a heavy template may be faster to rebuild than to trim.
If you want the number for your own homepage, the free site check includes it. The other things worth checking are in what a local business homepage needs. What to change first is in getting more calls from your website.
Sources
- Our own test of 212 Twin Cities roofing homepages, November 2026: what we found and how we tested.
- web.dev, Largest Contentful Paint and Web Vitals.
- Google for Developers, About PageSpeed Insights.
