←BaldanAll skillsLet's talk

Web, SaaS & Mobile · Web lab

Poke the web while it runs.

Four experiments that run in your browser, on this very page: how fast it feels by Google's measures, what it downloaded, how different rendering strategies would change that, and how a layout adapts to any screen.

01 · Live Web Vitals

How fast this page feels, by Google's measures.

Core Web Vitals are what Google uses to judge page experience: loading (LCP), responsiveness (INP) and visual stability (CLS). These are yours, measured as you use the page, and you can make them worse on purpose.

LCP

Largest Contentful Paint

…

not reported yet

when the biggest thing on screen finished painting

good ≤ 2.50 s · poor > 4.00 s

INP

Interaction to Next Paint

click!

no interaction yet

how long the page takes to react to your clicks and taps

good ≤ 200 ms · poor > 500 ms

CLS

Cumulative Layout Shift

…

not reported yet

how much things jump around while you read

good ≤ 0.100 · poor > 0.250

FCP

First Contentful Paint

…

not reported yet

when the first text or image appeared

good ≤ 1.80 s · poor > 3.00 s

TTFB

Time to First Byte

…

not reported yet

how long the server took to start answering

good ≤ 800 ms · poor > 1.80 s

Break it on purpose

Make the layout jump

A banner appears a second later and pushes the cards on the right down, like a late-loading ad. Watch CLS.

Freeze the page

This button runs 300 ms of heavy JavaScript before the page can repaint. Compare it with the fast one, and watch INP.

click anything to measure an interaction

Test stage

Article

Product

Price

Reviews

Measured in your browser with the same signals Google's web-vitals library uses; thresholds are Google's. Values are for this page visit only and can't be reset without reloading. Safari and Firefox don't report every metric, so some may show n/a.

02 · The waterfall

Every file this page loaded, in order.

The same view developers open in DevTools: when each file started, how long it waited for the server, how long it downloaded, and whether it came from your cache. Click any row.

…

files, including this page

…

transferred over the network

0

straight from your cache

…

slowest file

file00111

Reading this page's files…

queue, DNS, connect waiting for the server downloading (colour = file type) phases hidden by the other server

Read from your browser's Resource Timing after the page loaded; the time axis is milliseconds since you opened it. Files from other servers hide their phases unless those servers allow it.

03 · Rendering race

Same page, four ways to render it.

Where the HTML is built changes what people see first and when they can click. Pick a network, a slow or fast server and a small or heavy JavaScript bundle, then race them.

Network

round trip 70 ms · 1.5 MB/s

1.51 s
Buyclickable

Client-side (SPA)

an empty page, then JavaScript builds everything

connectempty HTMLfetch data
download JSrun JS
◆ first paint 949 ms★ content 1.42 s✓ clickable 1.42 sdata: live
Buyclickable

Server-rendered

the server fetches data and sends finished HTML

connectserver worksHTML
download JShydrate
◆ first paint 630 ms★ content 630 ms✓ clickable 1.37 sdata: live
Buyclickable

Static (pre-built)

HTML built at deploy time, served from a nearby CDN

connect (CDN)HTML
download JShydrate
◆ first paint 94 ms★ content 94 ms✓ clickable 785 msdata: as of the last build
Buyclickable

Streaming SSR

a skeleton at once, content streams in when the data is ready

connectshelldata streams in
download JShydrate
◆ first paint 198 ms★ content 630 ms✓ clickable 934 msdata: live
connectingHTMLserver fetching databrowser fetching datadownloading JavaScriptrunning JavaScript / hydrating

A model, not a measurement: two round trips to connect, the HTML is 30 KB, JavaScript costs 1 ms per KB to run on a mid-range phone, and a CDN edge is 35% as far away as the server. Green marks the winner of each moment. Try slow 3G with a big bundle, then a slow server: the winner changes. This site is pre-built (static) and fetches live numbers in the browser.

04 · Responsive playground

One page, every width.

A small shop page built with real CSS container queries. Drag its width from a small phone to a laptop and watch the layout switch exactly at its breakpoints.

390 px
Toko Kopi☰

Fresh Indonesian coffee, delivered.

Roasted this week in Makassar.

Kopi Toraja

Rp 85.000

Kopi Gayo

Rp 92.000

Kopi Kintamani

Rp 78.000

Cold brew

Rp 45.000

Drip bag ×10

Rp 60.000

Grinder

Rp 350.000

Breakpoints in this page's CSS

○@lg · 512 pxproducts in 2 columns
○@2xl · 672 pxmenu links appear, the ☰ button hides
○@3xl · 768 px3 columns and a bigger headline
○@4xl · 896 pxa filter sidebar appears

The classes doing it

<div class="@container">
  @lg:grid-cols-2
  @2xl:flex  @2xl:hidden
  @3xl:grid-cols-3  @3xl:text-[34px]
  @4xl:grid-cols-[170px_1fr]  @4xl:block

At 390 px only the mobile-first defaults apply. Container queries respond to the box, not the window, so the same component adapts wherever it's placed.

Drag the black handle or use the slider. On a phone the frame can only shrink to your screen; open this on a laptop to see every layout.