Page views
4.9k
30d
▲ 445% vs previous 30d
LCP · p75
6.13 s
27% good · 4.9k samples
▲ 4% vs previous 30d
INP · p75
220 ms
72% good · 3.2k samples
▲ 3% vs previous 30d
CLS · p75
0.03
89% good · 4.9k samples
▼ 2% vs previous 30d
FCP · p75
3.97 s
31% good · 4.9k samples
▲ 3% vs previous 30d
TTFB · p75
2.09 s
25% good · 4.9k samples
▲ 1% vs previous 30d
Happy page views
20%
51% unhappy · 4.9k rated
11% of these page views were in-app (soft) navigations. Those report INP and CLS, but browsers only measure LCP, FCP and TTFB on full page loads, so those three cover fewer page views.
What to fix first
Slow page loads cost you engagement
Comparing like with like (the same page type), page views whose main content took over 4 s got 54% less engaged time and 78% fewer clicks than ones that loaded in under 2.5 s. Based on 2354 slow page views across 5 page types.
Compare fast and slow page views →
/articles/* pages are slow to show their main content
p75 LCP is 6.39 s over 3013 full page loads; 50% took over 4 s. Most of the time goes on the image starting to load late.
Try: The browser finds the LCP image late: don't lazy-load it, give it fetchpriority="high", and make it discoverable in the HTML (or preload it).
LCP by page and element →
The server is slow to respond on /articles/*
p75 time to first byte is 1.97 s over 3013 page loads, so nothing can appear before then. Most of the time goes on waiting for the server.
Try: The server takes long to start sending the page: cache the HTML (at the CDN if you can) or speed up the code and database queries behind it.
TTFB breakdown →
People arriving from search get a slow first page
p75 LCP of the landing page is 6.19 s for 1445 search visits, and 39% of them bounced. It's their first impression, and page experience is one of Google's ranking signals.
Landing pages →
/ pages are slow to show their main content
p75 LCP is 6.42 s over 507 full page loads; 52% took over 4 s. Most of the time goes on the image starting to load late.
Try: The browser finds the LCP image late: don't lazy-load it, give it fetchpriority="high", and make it discoverable in the HTML (or preload it).
LCP by page and element →
The server is slow to respond on /
p75 time to first byte is 2.11 s over 507 page loads, so nothing can appear before then. Most of the time goes on waiting for the server.
Try: The server takes long to start sending the page: cache the HTML (at the CDN if you can) or speed up the code and database queries behind it.
TTFB breakdown →
11 more findings
/search/ pages are slow to show their main content
p75 LCP is 5.87 s over 488 full page loads; 48% took over 4 s. Most of the time goes on waiting for the server.
Try: Most of the time passes before the HTML arrives: cache pages at the edge/CDN or speed up the server.
LCP by page and element →
/p/* pages are slow to show their main content
p75 LCP is 6.35 s over 453 full page loads; 54% took over 4 s. Most of the time goes on the image starting to load late.
Try: The browser finds the LCP image late: don't lazy-load it, give it fetchpriority="high", and make it discoverable in the HTML (or preload it).
LCP by page and element →
The server is slow to respond on /p/*
p75 time to first byte is 2.00 s over 453 page loads, so nothing can appear before then. Most of the time goes on waiting for the server.
Try: The server takes long to start sending the page: cache the HTML (at the CDN if you can) or speed up the code and database queries behind it.
TTFB breakdown →
Mobile visitors get a much slower INP
p75 INP is 272 ms on mobile against 141 ms on desktop. Mobile is 51% of your page views.
Try: Test on a real mid-range phone or with CPU throttling: big images, heavy JavaScript and third-party scripts cost far more there.
INP by device →
Taps and clicks on this element respond slowly
#onetrust-accept-btn-handler
p75 INP is 222 ms over 779 page views, mostly on /. Most of the time goes on running the event handler.
Try: The event handler itself is slow: do less work in it, or yield (scheduler.yield / setTimeout) before the heavy part.
INP by element →
/c/* pages are slow to show their main content
p75 LCP is 3.63 s over 481 full page loads; 21% took over 4 s. Most of the time goes on waiting for the server.
Try: Most of the time passes before the HTML arrives: cache pages at the edge/CDN or speed up the server.
LCP by page and element →
This element makes the page jump as it loads
header.site-header
p75 CLS is 0.10 on the 500 page views where it shifted, mostly on /.
Try: Reserve its space before it arrives: width and height (or aspect-ratio) on images and embeds, a min-height for content inserted later, and font-display with matched fallback metrics for web fonts.
CLS by element →
This element makes the page jump as it loads
div.ad-slot
p75 CLS is 0.10 on the 510 page views where it shifted, mostly on /search/.
Try: Reserve its space before it arrives: width and height (or aspect-ratio) on images and embeds, a min-height for content inserted later, and font-display with matched fallback metrics for web fonts.
CLS by element →
This element makes the page jump as it loads
ul.product-grid
p75 CLS is 0.10 on the 484 page views where it shifted, mostly on /search/. 1 more element that shifts is in the Vitals tab.
Try: Reserve its space before it arrives: width and height (or aspect-ratio) on images and embeds, a min-height for content inserted later, and font-display with matched fallback metrics for web fonts.
CLS by element →
Taps and clicks on this element respond slowly
nav.toc > a.toc-link
p75 INP is 213 ms over 454 page views, mostly on /. Most of the time goes on running the event handler.
Try: The event handler itself is slow: do less work in it, or yield (scheduler.yield / setTimeout) before the heavy part.
INP by element →
Taps and clicks on this element respond slowly
div.gallery > button.next
p75 INP is 249 ms over 388 page views, mostly on /search/. Most of the time goes on running the event handler. 4 more slow elements are in the Vitals tab.
Try: The event handler itself is slow: do less work in it, or yield (scheduler.yield / setTimeout) before the heavy part.
INP by element →
Trend
By device
| Device | Page views | p75 LCP | p75 INP | p75 CLS |
|---|---|---|---|---|
| Mobile | 2.5k | 6.85 s | 272 ms | 0.03 |
| Desktop | 1.8k | 4.70 s | 141 ms | 0.03 |
| Tablet | 618 | 7.30 s | 260 ms | 0.04 |
Does speed cost engagement?
Page views grouped by their own LCP. If slower pages get fewer clicks and less reading, that's the cost of slowness.
| LCP | Page views | Clicks per view | Engaged time | Avg scroll |
|---|---|---|---|---|
| Good | 1.4k | 1.68 | 34.7 s | 57% |
| Needs improvement | 1.2k | 0.87 | 24.3 s | 57% |
| Poor | 2.4k | 0.38 | 15.9 s | 38% |
Biggest changes vs previous 30d
| Measure | Previous | Now | Change |
|---|---|---|---|
| p75 LCP · Load time | 1.58 s | 1.65 s | ▲ 4% |
| p75 LCP | 5.92 s | 6.13 s | ▲ 4% |
| p75 FCP | 3.85 s | 3.97 s | ▲ 3% |
| p75 LCP · Load delay | 3.69 s | 3.79 s | ▲ 3% |