Jetpack Boost- Does it work?

In order to help WordPress websites achieve the Core Web Vitals scores, Jetpack released the Jetpack Boost plugin. Did it help?

TL;DR: it helped the CWVs and broke the website.

Here’s the before and after data:

I’d love to hear other peoples’ experiences with Jetpack Boost. I only ran this on one site which makes these results anecdotal. Send me your data!

Transcript:

Jetpack Boost. Did it work?

Okay, Jetpack Boost. So last time we talked about this new WordPress plugin called yet back Jetpack Boost that was supposed to fix the three web core vitals, which will become a ranking signal in June through August.
I installed it on The client’s website and immediately broke it. And, in hindsight, of course, it was going to, because to fix the things that it says, it has to fix it. It had to remove something and remove the code.
I made a baseline comparison of data. Let me see if I meant actually publish that.

Is this why everybody all of a sudden is talking about page speed?

Oh yes. Oh yeah, exactly.

I’m going to share the results with you: before and after. The first line here is the performance score. This is the page speed score between 0 and 100. The second is the first of the web core vital’s largest contentful paint. The third is the closest we can measure to the second web core vital is not the first input delay. Still, it’s as close as we can measure objectively called time to interactive. And third is content layout shift. This is before I added the plugin. This is with the plugin (I had to immediately remove the plugin).

The average score before the plugin on this website was 60—the median 68. So people who aren’t math people average is the total divided by the number, you know, the median is the most common. So if there’s a difference between average and median, the median means that there’s more higher than lower. 68 is a yellow score, not bad. But not terrible. The worst page before the plugin was a 33 out of a hundred- that’s considered red. The best page before the plugin was 86. That’s pretty good, still yellow. If we just did a performance score with this plugin, we increased our average from 65 to almost 80. The median being (meaning, the most common) is 92. That’s pretty darn good. However, the worst was now 16. One of them went from 30- well, we don’t know if it’s the same one that went from- 33 to 16. Now there’s a 16 out of a hundred, whereas before there’s only a 33 was the worst score. However, the best page is 99, so that’s really good.

We can do the same thing with the largest contentful paint (these are in milliseconds). So, we went from five seconds average (4.8 seconds, median)- still yellow. So, this did not do a lot to fix the largest contentful paint. In other words, the lowest largest contentful paint time was 1.8 seconds, which is a green score. But gosh, one was still 10 seconds. So that could mean like there was a massive image on that page.

Time to interactive. I didn’t really compare that, but you can kind of see some ideas there.

Commutative layout shift, which is one of the hardest things to do. It’s not a speed thing as much as a codes written thing. This plugin really did improve it significantly. Meaning that some pages had a zero, which is optimal- like perfect for cumulative layout shift.

So the plugin worked. If the only thing was to get these three scores, this would have improved the site- except for this weird thing where the performance score on some page was actually lower. I don’t understand that.
But it totally broke the site. It was so bad that while I was preparing this data, the client emailed me, “David, It’s broken.” I got to take this down quick! It just so happened I did it on, like, a Friday morning.

So it looks like the data shows that the Jetpack Boost does help but caution- buyer, beware (buyer, beware? it’s free!). It could break your site. So, I would recommend if you install it, you do it during a low-traffic time, maybe like in the evening Friday evening because there’s nothing more fun to do than spend your Friday evening installing WordPress plugins (says, nobody ever).

Or Sunday morning.

Oh, yeah. Sunday morning.

There you go. Here’s some data for you. I hope this helps. I should have zoomed this out so you can see it a little bit better. Anyway, clearly makes an improvement. But yeah, it can break your website.

So is this plugin still in what they consider a beta state, or, you know, are they really saying?

It’s ready for market as of last week- they said it was prepared for market. We installed it on this client’s site the week before when they released it.

I have to do with this client as I got to go through and make sure there are no of these super large images. Which I expect there are. I fact, I know there are. The homepage was totally broken. So we couldn’t use this on this site to solve these problems. So, what this client is looking at is a complete website redesign at this point.

But I’d love to see anybody else’s data. In fact, if you- I will even open this up to you to this group- if you want to try it, I will run a Screaming Frog crawl of your website- before and after and come up with this data.

Is that a technical term: “screaming frog crawl?”

Yeah, hmm. Yeah. It’s just a little frog that screams all the way through your website. It’s really cool. How does it sound, Trisha?

I can’t remember. I, it’s a bizarre sound.

That’s what I’m sure you’ve heard. You could do it. It’s a great tool that crawls websites for you.

That is bizarre.

It’s a great tool. It’s one of my favorite tools. But Lavonya, does that answer your question?

Yes, it does.

I don’t know if it’s conclusive, but there’s some data.


Have a question about this process? Ask it here: