Understanding Google Site Performance and Web Vitals

Posted by Owen Powis on 9 Sep, 2020
View comments SEO
Even if you think it’s good, trust me when I say, it’s time to recheck your site’s performance according to Google.

We engineered our website to perform well on mobile and desktop, so I was a little shocked to see that our performance scores have plummeted across Mobile performance according to Google’s Page Speed Insights tool.

Of course we keep an eye on page speed load times, I mean there’s a whole tab for that in Google Analytics.

Navigate to Behaviour > Site Speed > Overview to find yours. Year on Year analysis shows we are performing better than before:

Adding a Segment is a quick and easy way to grab a different view on that metric so, to check these numbers for Mobile just use the Add Segment section at the top:

Uncheck ‘All Users’ and select ‘Mobile Traffic’ and you’ll quickly change the data to be Mobile only.

We can see a small difference in Server Response Time but nothing dramatic and the actual Page Download Time has decreased so we should be looking good.

However looking within Google Page Speed Insights tool shows us another story:

Yeah, ouch. This page was previously in the high 40’s to low 50’s which is pretty good as our homepage has some heavy lifting to do. Checking some more lightweight pages, like from the Blog shows a better but not exactly fantastic score either:

However when we check the Desktop scores for these pages:

Nothing to write home about, but a pretty big disparity between the two. So we know our mobile scores have plummeted yet our reported performance in GA hasn’t.

The thing is it’s not just us. I checked a bunch of competitors and I was going to post a comparison table then felt it better not to name and shame. But, it’s more a case of doing less bad than doing well (scores ranged from 5 - 30 with one exception and largely depended on page type).

Oh and in case you were wondering…

I tried this as did my colleagues on multiple machines and connections. I mean I’m sure it’s just an, ahem coincidental error, cough, so let me know in the comments if it works for you :)

Also from all the testing I did, I have to give a big shout out to Barry and the guys over at SERoundtable.com as they are absolutely killing it:

This was the best Mobile performance score I saw on any property, including Google Blog's own score (50). So whilst I won't highlight the worst (it wasn't us!) I will say these guys are doing the best out of anyone I tested.

But, it's also worth bearing in mind it also depends on the type of page.

So what the hell, Google?

How are we all failing so badly (except Barry!), especially when I know we used to perform significantly better?

To understand that, it's worth knowing how the Page Speed Insights tool works and what information Google is trying to give us with it.

The tool is broken down into two core components, and fundamentally different ways of collecting data. One is the ‘Field Report’ which if available (low traffic pages may be out of luck) shows data collected from the Chrome User Experience Report, or as it is referred to CrUX.

The CrUX feeds anonymised performance data back to Google from Chrome users and this is shown in the report over a 28-day average. Google will also run its own tool, Lighthouse, to collect Lab Data about the page. This is Google’s attempt at simulating a static set of network conditions to give a consistent report. There is some variance in this depending on when / where you check it, but it’s generally pretty consistent.

Within the Field and Lab data some of the core metrics are highlighted with a blue flag icon. These are the metrics which are counted as part of Google’s new Core Web Vitals assessment.

These are the metrics that Google is going to be concentrating more on moving forwards and focus on "loadinginteractivity, and visual stability". We know that Google was planning on including these within the algorithm directly, but have pushed back that change citing Covid-19 and giving webmasters at least 6 months' notice:

A note on timing: We recognize many site owners are rightfully placing their focus on responding to the effects of COVID-19. The ranking changes described in this post will not happen before next year, and we will provide at least six months notice before they’re rolled out. We're providing the tools now to get you started (and because site owners have consistently requested to know about ranking changes as early as possible), but there is no immediate need to take action.

https://webmasters.googleblog.com/2020/05/evaluating-page-experience.html

We also know this is going to be a ranking factor that is likely going to sit mid pack for Google, from Gary Ilyes' comments on Reddit on the issue:

Like any other search engine, Google works hard to surface the highest quality and most relevant results for users' queries. CWV has nothing to do with either of those, not even remotely, so it's extremely unlikely that CWV would ever become "the primary factor for Organic Traffic".

That's not to say you can ignore CWV, though.

https://www.reddit.com/r/TechSEO/comments/innc2c/will_cwv_core_web_vitals_be_the_primary_factor/

If you’re not on that Sub Reddit I'd definitely advise joining or at least ‘lurking’ (those that read the posts without joining) every now and again.

So what can I do to sort this out?

We're not going to immediately rip apart and re-engineer our entire site for site speed. Re-engineering the homepage would be a massive undertaking, and something that I never see SEO's talk about enough is cost / benefit. That's my byword to our clients and I think it needs to be something that's far higher up on the list of any recomendations. Prioritising in terms of SEO impact is easy. But sitting down with a client, understanding their infrastructure, staffing and in-house technical ability and budget is where you really earn your fee.

Tailoring the approach according to what's achievable is the route to success and for us that's not going to be a homepage redesign to fix this issue. However our homepage ranks well and it's not where I believe Google values mobile usability the most. That would be on our news content - meaning our Blog and Academy articles. That's the content that users want to be able to read on the train on the way to work, sat on the sofa in the evening and on the go. So that's the content we can concentrate on making as mobile friendly as possible.

For us that hugely simplifies the task as our blog is on a seperate framework on the backend to our homepage. So we're going to focus on what we can do to reduce the load times in those areas and bring those scores up. Also, that content is simpler, with fewer technical requirements such as interactive elements.

Luckily enough Google gives lots of guidance on what you can do to improve each of the individual scores:

https://web.dev/lighthouse-performance/

Before you do this you may want to want to check out the Lighthouse Scoring Calculator. This allows you to change the values for the different metrics they measure to see how you would score with better values:

So, if you're wondering which values to start with, look at what is most easily achievable and balance that against the impact of lowering the value in the calculator. LCP (Largest Contentful Paint) has the joint highest weighting, signifying that Google views it as one of the most meaningful, so that's usually a good place to start looking. In addition, optimisations in this area will usually benefit a few of the other metrics.

We'll be looking into these metrics in more detail in future articles so look out for further technical content in this area from us.

Recent articles

Google Search ranking bug affecting large numbers of results
Posted by Edith MacLeod on 18 August 2024
Google releases August 2024 core update
Posted by Edith MacLeod on 16 August 2024
Google updates trending now report in Trends
Posted by Edith MacLeod on 13 August 2024
How to identify and avoid SEO vanity metrics
Posted by Syed Balkhi on 12 August 2024
Google launches SEO recommendations
Posted by Edith MacLeod on 6 August 2024