Sample report · technical measurement · September 2026
This site appears within two seconds. It only starts listening after six.
This is a genuine technical report, with the client’s name removed. A judgment is provided for each finding: replace, repair, keep as is, or withdrawn. The latter is included intentionally.
The same homepage, the same day
Desktop · broadband93
Mobile · 4G65
GTmetrix performance score on the homepage, both on September 12, 2026. Chrome 142, Lighthouse 12.6.1. Mobile on an iPhone 13 profile with 4G (9/5 Mbps, 125 ms), desktop on broadband (20/5 Mbps, 25 ms). Name, address, and page names have been omitted; figures have not been adjusted.
The core in four numbers
Interactive
6.0 s
until a tap does something, averaged over nine pages on 4G; standard 2.5 s
Slowest page
12.6 s
until interactive; the fastest of the nine reaches 2.7 s
Third-party
63 %
of the JavaScript on the homepage: tag manager, heatmaps, pixels
Duplicate H1
441
pages with two main titles, sitemap and service pages combined
The homepageHomepage The first page of a website, which you see when you only type in the domain name. achieves an A on desktop. Anyone who only measures that will miss the problem. The pages a customer visits to be persuaded are two to six times slower than the homepage. On a phone, the site only responds to a tap after more than four seconds.
Five components, each with a judgment
Speed: nine pages, one connection
Nine pages, measured within ten minutes of each other on the same 4G profile. Interactive is the moment the page responds to a tap; finished is when it stops loading and moving. Not a single page meets the 2.5-second standard.
Seconds until interactive and until finished, nine pages on 4G
Black: current time to interactive, gray: time to finished. Blue: the same page, rebuilt by EQwise, same measurement. The dotted line is the 2.5-second standard.
| Page | Judgment | Interactive (s) | Finished (s) | Weight (kB) | Requests |
|---|---|---|---|---|---|
| Reviews | replace | 12,6 | 15,9 | 11.124 | 188 |
| References | replace | 7,6 | 8,9 | 3.004 | 188 |
| Service A | replace | 7,3 | 11,3 | 5.892 | 172 |
| Service B | replace | 6,3 | 8,3 | 3.314 | 160 |
| Service C | replace | 5,9 | 8,3 | 4.693 | 215 |
| Articles | repair | 5,0 | 7,6 | 3.309 | 181 |
| Service D | repair | 3,7 | 5,7 | 1.653 | 134 |
| About us | repair | 3,1 | 6,4 | 1.525 | 136 |
| Service E | repair | 2,7 | 6,2 | 2.091 | 142 |
- Reviews: downsize images and lazy load
- References: same recipe
- Service A: nearly 7 MB of imagery on a page where someone is considering an expensive service
Where the weight lies.
The reviews page is nine-tenths imagery: 9,907 kB in 69 files, for 49 customer quotes. Additionally, 745 kB of JavaScriptJavaScript A programming language that makes web pages interactive, such as a drop-down menu. in 86 files and 201 kB of fonts in 6 files. The same page, rebuilt without a page builderPage builder An additional program in WordPress that allows you to build pages by dragging and dropping components. and with custom-sized imagery, weighs 586 kB in 17 requests and is interactive after 1.0 second. That is a measurement, not an estimate.
Google’s three metrics
The metrics Google uses itself, on the homepage, on a phone, and on a desktop. The standard is Google’s: largest element in view within 2.5 seconds, blocking timeBlocking time The time during loading when a page does not respond because it is processing code. under 200 milliseconds, layoutTemplate A fixed layout used for an entire group of pages simultaneously. shift under 0.1.
| Metric | Mobile, 4G | Desktop | Standard | What it means |
|---|---|---|---|---|
| Largest Contentful PaintLCP The time it takes for the largest visible element of a page to appear on the screen. | 2.2 s pass | 1.3 s pass | ≤ 2.5 s | The page appears on time. This is not the problem. |
| Blocking time | 259 ms fail | 2 ms pass | ≤ 200 ms | On a phone, there are moments when a tap does nothing: the page is busy processing scripts. Nine days earlier, this was 6 ms; the site is growing. |
| Layout shift | 0.09 marginal | 0 pass | ≤ 0.1 | Images without dimensions only take up space once they arrive; the text shifts while you are reading. |
| Fully loaded | 6.3 s | 3.0 s | guideline 3 s | On the limit for desktop, well over it on a phone. |
What Google knows of the site
Measured on September 8 and again on September 15, with only status codesStatus Code A three-digit code used by a server to indicate whether a page has been found. and the sitemapSitemap A list of all your website's pages, specifically designed for search engines.. One finding was resolved by the second measurement and is intentionally still included.
| Finding | Number | Judgment |
|---|---|---|
| Pages in the sitemap, with last modified date | 201 | keep as is the sitemap is correct and maintained |
| Pages with more than one main titleH1 The main title of a page, as defined within the code., sitemap and service pages combined | 441 | repair the title appears in the top bar and in the text, both as main titles; Google and AIAI Computer programs that perform tasks normally requiring human intelligence, such as writing text or answering questions. answers then have to guess what the page is about. One line in the template |
| Service cards arriving via a redirect | 13 of 35 | repair seven services have been renamed, six cards are missing a slash; every click costs an extra round trip to the server |
| Two different services at the same address | 1 pair | repair one of the two is unfindable for those who click on it |
| Service cards to a page that does not exist | 5 → 0 | withdrawn still five on September 8, resolved by the client themselves by September 15. We remove findings from the report once they are resolved |
| Robots, canonicals, structured data | — | keep as is no blockages found |
On a mobile phone
The same homepage, the same day, a phone versus a desktop. The difference isn’t in the imagery but in the processing power the page demands from a phone. Think of 919 elements that the browser recalculates with every movement and 259 milliseconds where a tap does nothing. Additionally, there are seven separate files for a single font.
- Images arrive at desktop size and are downsized by the phone itself. It is an older format, often two to four times larger than the slot they occupy. Judgment: replace with a modern format sized to the screen.
- Nothing falls off the screen at 375 and 390 px, and buttons are large enough for a thumb. Judgment: keep as is.
- The page only responds to a tap after more than four seconds, even though it is readable after two. Visitors tap in the meantime and think the site is broken. Judgment: only load third-party scripts after the initial render.
What a visitor doesn’t see but does feel
These five together explain why a faster server wouldn’t help.
| Finding | Number | Judgment |
|---|---|---|
| Dependency chains: files waiting for each other like a checkout line | 99 chains, 137 requests | replace one central source for styling and code without intermediate steps. A rebuilt page has two |
| Third-party scripts before initial render: tag manager 345 kB, heatmaps 65, ad pixel 40, pop-ups 18, CRM 3 | 471 kB · 63% of all JavaScript | repair load after consent and after initial render. This is a marketing choice, not a fundamental problem: no rebuild will remove it |
| Fluctuating server time: sometimes the page is ready, sometimes it is generated on the fly | factor of 5 | repair a stock of ready-made pages that stays warm |
| The same page, twice as slow: three desktop measurements with everything identical gave 1.7 s and 3.9 s | factor of 2.3 and 34 | repair same cause; continuous monitoring makes it visible the day it happens |
| The site grows without the content changing: between September 3 and 12, 3 requests and 60 kB were added | +60 kB in 9 days | repair a fixed measurement setup that runs monthly |
What meets the standard
You hear what is good just as much as what isn’t; that determines what you can leave alone.
- On desktop, the homepage meets all three Google metrics: 1.3 s, 2 ms, 0.
- The sitemap is complete and up to date, and the content is maintained.
- Redirects were set up when renaming services; broken links are gone.
- Nothing falls off the screen on a phone, and tap targets are large enough.
- No blockages for search engines.
What this does and does not prove
What it proves
- These nine pages only become interactive after 2.7 to 12.6 seconds on 4G. That is due to the build (chains, page builder, imagery) and not the server.
- The same page without a page builder and with custom-sized imagery is interactive after one second: measured across nine rebuilt pages, 0.9 to 1.0 s.
What it does not prove
- That every visitor experiences exactly these figures; that depends on the device and connection. Visitor data from Search Console is needed for that.
- That the entire site lands at one second: pages with video or a checkout step weigh 200 to 300 kB more.
How it was measured
A service provider with over two hundred pages on WordPress, using a page builder, active in an English-speaking market. Eighteen GTmetrix measurements between September 3 and 13, 2026: Amsterdam and London, Chrome 142, Lighthouse 12.6.1. Testing was conducted with an iPhone 13 profile on 4G (9/5 Mbps, 125 ms) and a desktop on broadband (20/5 Mbps, 25 ms).
Status codes and sitemap on September 8 and 15. This is an anonymized example of a real measurement: the name, web address, and page names have been omitted or replaced by a description.
How fast is your site on a phone?
The front door is almost always faster than the shop behind it. On this site, that ranged from 2.7 to 12.6 seconds before a visitor could do anything, on the same connection and the same day. You don’t see that difference in a single measurement of the homepage.
