I think the same about it as I did with the previous Google Pagespeed Insights.
- Has some useful metrics to help you optimize your site for performance, SEO, performance, accessibility, and proper coding standards.
- But also has some misleading or inaccurate metrics.
- Not a helpful overall context of asset load and what things are going right.
Try measuring your site at https://web.dev/measure/ and let’s go over what we see.
Overall, I’m not a fan.
- This is the Web Vitals report for wpjohnny.com
If you like it and it helps you, by all means keep using it. But for my personal workflow, it’s a useless tool. Here’s why…
1. Misleading or inaccurate metrics.
Look at the report for my site. Time to interactive is 4 seconds? And largest contentful paint is nearly 5 seconds? That’s just wrong.
And I get why. It’s because Web Vitals, as with any other speed testing tool, can’t accurately measure for every possible type of web application. WordPress loads differently, Joomla loads differently, Shopify loads differently, every CMS is different.
Imagine testing different kinds of cars. Maybe in one model, a certain type of sound is good. In another model, that same sound could be bad. Anyway…moving on.
2. Misleading priorities
This is a problem with any page testing tool, really. (I’ve covered it before in my long rant about why page scores don’t matter). They prioritize certain metrics in ways that don’t actually reflect your site load.
If you don’t know what you’re doing…you’re going to freak about all the red items for no reason. They’ll make you think THAT is why your site is slow.
But more often than not, your site is slow for some other reason…and quite often, that reason isn’t even covered in the page tool!
So basically, they make a big stink out of little things, while still missing the big things. And of course, most people won’t know any better. They’ll argue with their developers (ugggghhhhh).
3. Lacking helpful overview
Web Vitals doesn’t give you an overview of how your site loads and where things are. There’s no waterfall, no list of assets. Instead it just gives you a list of automated recommendations.
I’m sorry but that’s not very useful to professionals.
Professionals simply prefer to see numbers and facts and we decide how we want to interpret them. We don’t need some automated tool to take random guesses at what’s good or bad.
Is Google Web Vitals totally useless?
I personally wouldn’t use it at all. Wouldn’t let it dictate my development or speed optimization work whatsoever. The metrics are not useful to how I work.
Does it mean that I shouldn’t care about any of their metrics?
- Of course not. I do care but I measure them in a different way. Not only that but personally, I already measure several other things that are more accurate to my clients pages and application software.
- Again, you have to remember that Web Vitals doesn’t specifically test for only WordPress sites. It tests for all sites. Would you apply a generic checklist for Toyota cars to your Ford truck? Of course not!
Does it mean you can’t use the tool?
- You can do whatever you want. If you find it useful and you like the results. Go ahead.
Will a site be penalized if it doesn’t satisfy all the recommendations?
- No (not in my experience). Many super-fast high-ranking sites don’t satisfy all those recommendations.
- But you’re welcome to follow your fears anyway you want.
Will getting higher page scores help differentiate ranking between 2 similar sites?
- In my experience, no. But you’re welcome to do your own independent testing.
- Try searching up keywords for any niche and compare their SEO ranking with page scores.
What should most users do? Follow or don’t follow Web Vitals?
- Most users aren’t experienced enough to interpret these tools correctly. Whether they follow or don’t follow the tool won’t help them much either way IMO (for performance or SEO).
- They should be using developers who KNOW which metrics are important to which sites.
What tool does Johnny prefer then?
- GTmetrix is my favorite. But that doesn’t mean I freak out and care about all the metrics it gives. I only pay attention to a few certain places.
- Last but not least, I optimize my (and client’s) websites for best load and UX! I optimize for humans, not test scores.
Any thoughts on https://www.fastorslow.com , WordFence new service?
I actually really like FastorSlow. It’s a mix of GTmetrix, Pingdom, and WonderNetwork. The numbers are more unbiased and not overly interpreted. It simply gives you the numbers and lets you read them for yourself. Really great tool and no surprise at all if you remembered Wordfence’s history in caching.
Thanks for your thoughts, just wanted to confirm my own thoughts, and you did .. =)
Daan van den Bergh
I read somewhere that Core Web Vitals are going to part of Google’s ranking strategy in about 6 months.
I don’t care about the overall Performance score. It’s a cool gimmick, nothing more. But getting the Core Web Vitals in the green, that’s definitely something to aim at, SEO-wise. Regardless of if they’re reasonable, accurate or even worth something.
I personally don’t worry about that stuff because my work already satisfies that and more.
But when it comes to clients with poorly-built pages, it’s always a trade-off. Either satisfy that at the cost of other important conversion-tools and services…or ignore. Many clients don’t know and make poor decisions that only satisfy unimportant things while hurting UX, overall load time, and even SEO and conversion rate in the end.
Ultimately, if you know what you’re doing…you can do what you want. And if you don’t know what you’re doing….following simplified recommendations won’t improve your site at all. If you have a better way of phrasing it, let me know.
In the meanwhile, I face an uphill battle with naive clients arguing it should be possible to change expiry time on externally-loaded assets. They argue that Google recommends it, so Google must be right.
This article confirms my own thoughts about web.dev
I have a telemetry plugin in my browser which measures time to load 0.6s to my own website hosted on a VPS I manage, in a local datacenter with 40Gbps link to the main IX in my country (South East Europe).
Google measures my speed by opening my side from California, literally at the other end of the world, and tells me they test with an emulated device running Android 7 and throttled at 1.6Mbps down / 750 Kbps up.
My customers use iPhones and Samsung Galaxies. I know, I see the user agent stats. We pride with having three mobile telecoms that provide 4G LTE with up to 112/37.5Mbps speed. I regularly hit 80 Mbps on a mid-tier phone. How does Google measuring performance like this make any sense?
As for what tools I use, I dislike GTmetrix’s stupid insistence on punishing me for not having cookie-free resources (Cloudflare doesn’t allow this… sorry) or giving too much weight to having a CDN, yet failing to recognize CF as such. So while I use GTmetrix, I prefer webpagetest.org for the much larger choice of test locations and options to throttle speed to simulate.
Loving this blog! Thank you for all the valuable info Johnny is sharing with the WP community.