Learnings From a WebPageTest Session on CSS-Tips

No Comments

I obtained along with Tim Kadlec from over at WebPageTest the opposite day to make use of do a little bit of efficiency testing on CSS-Tips. Basically use the instrument, poke round, and determine efficiency ache factors to work on. You possibly can watch the video proper right here on the positioning, or over on their Twitch channel, which is price a subscribe for extra efficiency investigations like these.

Net efficiency work is twofold:

Step 1) Measure Issues & Discover Issues
Step 2) Repair it

Tim and I, by means of the wonderful instrument that’s WebPageTest, did a variety of Step 1. I took notes as we poked round. We discovered various drawback areas, some pretty massive! In fact, in spite of everything that, I couldn’t get them out of my head, so I needed to spring into motion and do the Step 2 stuff as quickly as I may, and I’m blissful to report I’ve finished most of it and seen enchancment. Let’s dig in!

Recognized Drawback #1) Poor LCP

Largest Contentful Paint (LCP) is among the Core Net Vitals (CWV), which everyone seems to be fastidiously watching proper now with Google telling us it’s an search engine optimisation issue. My LCP was clocking in at 3.993s which isn’t nice.

WebPageTest clearly tells you if there are issues along with your CWV.

I additionally discovered from time that it’s preferrred if the First Contentful Paint (FCP) accommodates the LCP. We may see that wasn’t occurring by means of WebPageTest.

Issues to repair:

Be sure the LCP space, which was finally a giant picture, is correctly optimized, has a responsive srcset, and is CDN hosted. All these issues have been failing on that exact picture despire working elsewhere. The LCP picture had loading=”lazy” on it, which we simply discovered isn’t a superb place for that.

Fixing method and learnings:

All the correct picture dealing with stuff was in place, however for no matter purpose, none of it really works for .gif recordsdata, which is what that picture was the day of the testing. We in all probability simply shouldn’t use .gif recordsdata for that space anyway.Flip off lazy loading of LCP picture. This can be a WordPress featured picture, so I basically needed to do <?php the_post_thumbnail(”, array(‘loading’ => ‘keen’)); ?>. If it was an inline picture, I’d do <img data-no-lazy=”1″ … /> which tells WordPress what it must know.

Recognized Drawback #2) First Byte to Begin Render hole

Tim noticed this instantly as a reasonably apparent drawback.

Within the waterfall above (right here’s an excellent detailed article on studying waterfalls from Matt Hobbs), you’ll be able to see the HTML arrives in about 0.5 seconds, however the begin of rendering (what folks see, massive inexperienced line), doesn’t begin till about 2.9 seconds. That’s too dang lengthy.

The chart additionally identifies the issue in a yellow line. I used to be linking out to a third-party CSS file, which then redirects to my very own CSS recordsdata that comprise customized fonts. That redirect prices time, and as we dug into, not simply first-page-load time, however each single web page load, even cached web page masses.

Issues to repair:

Remove the CSS file redirect.Self-host fonts.

Fixing method and learnings:

I’ve been eying up some new fonts anyway. I famous not way back that I actually love Mass-Driver’s licensing innovation (priced by # of staff), however I equally love MD Primer, so I purchased that. For physique sort, I caught with a snug serif with Blanco, which mercifully got here with very properly optimized RIBBI1 variations. Subsequent time I swear I’m gonna discover a variable font, however hey, you gotta observe your coronary heart typically. I bought these, and am now self-hosting the font-files.Use @font-face proper in my very own CSS, with no redirects. Additionally utilizing font-display: swap;, however gotta work a bit extra on that loading method. Can’t look forward to size-adjust.

After re-testing with the change in place, you’ll be able to see on a giant article web page the beginning render is a full 2 seconds sooner on a 4G connection:

That’s a biiiiiig change. Particularly because it impacts cached web page masses too.

See how the waterfall pulls again to the left with out the CSS redirect.

Recognized Drawback #3) CLS on the Grid Information is Dangerous

Tim had a neat trick up his sleeve for measuring Cumulative Format Shift (CLS) on pages. You possibly can instruct WebPageTest to scroll down the web page for you. That is vital for one thing like CLS, as a result of structure shifting may occur on account of scrolling.

See this text about CLS and WebPageTest.

The trick is utilizing a complicated setting to inject customized JavaScript into the web page in the course of the take a look at:

At this level, we have been testing not the homepage, however purposefully an important web page: our Full Information to Grid. With this in place, you’ll be able to see the CWV are in a lot worse form:

I don’t know what to assume precisely in regards to the LCP. That’s being triggered by what occurs to be the biggest picture fairly far down the web page.

I’m not terribly fearful in regards to the LCP with the scrolling in place. That’s just a few picture like another on the web page, lazily loaded.

The CLS is extra regarding, to me, as a result of any shifting structure is at all times obnoxious to customers. See all these dotted orange traces? That’s CLS occurring:

The orange CLS traces correlate with photos loading (because the web page scrolls down and the lazy loaded photos are available in).

Issues to repair:

CLS is unhealthy due to lazy loaded photos coming in and shifting the structure.

Fixing method and learnings:

I don’t know! All these photos are inline <img loading=”lazy” …> parts. I get that lazy loading may trigger CLS, however these photos have correct width and peak attributes, which is supposed to order the precise area mandatory for the picture (even when fluid, due to side ratio) even earlier than it masses. So… what provides? Is it as a result of they’re SVG?

If anybody does know, be happy to hit me up. Such is the character of efficiency work, I discover. It’s a mix of straightforward wins from foolish errors, little battles you’ll be able to struggle and win, greater battles that typically includes exterior influences which are more durable to win, and mysterious unknowns that it takes time to heal. Luckily now we have instruments like WebPageTest to inform us the true tales occurring on our web site and provides us the perception we have to struggle these efficiency battles.

RIBBI, I simply discovered, means Common, Italic, Daring, and Daring Italic. The traditional combo that almost all physique copy on the internet wants.

The submit Learnings From a WebPageTest Session on CSS-Tips appeared first on CSS-Tips. You possibly can assist CSS-Tips by being an MVP Supporter.

    About Marketing Solution Australia

    We are a digital marketing company with a focus on helping our customers achieve great results across several key areas.

    Request a free quote

    We offer professional SEO services that help websites increase their organic search score drastically in order to compete for the highest rankings even when it comes to highly competitive keywords.

    Subscribe to our newsletter!

    More from our blog

    See all posts

    Leave a Comment