CLS April Update

[Brad Billman]

Hello,

I run a team that has built an SPA with SSR on Vue2.  Search console showed that we had 99% good urls up until April 13th.  It dropped to 0 over the following days.  I've spent countless hours trying to understand the scores but I just realized today using the web vitals extension that our client side routing is causing big CLS events.  Client side routing obviously shifts the entire page because that is the point of it.

Most of our pages have good CLS.  I did find a couple of small improvments I can make, but I have no solution for the routing CLS.

It is a listing site where users will often go back and forth comparing different listings.  In my tests each new [client side] page load ads about 0.1 CLS and it can rack up to 0.5 or even 0.8 quickly.

All of this started with the new update.  Is there any general advice to keep client side routing from counting as CLS?

Thanks

[Philip Walton]

Thanks for the feedback! It's hard to know for sure what's causing the high CLS on your pages, but based on this comment:

> It is a listing site where users will often go back and forth comparing different listings.

My assumption is the issue is your SPA route changes are causing a shift which is being ignored (hadRecentInput = true) when initiated from the user by tapping a link, but not ignored (hadRecentInput = false) when initiated by tapping the back/forward button.

I checked with the team and we agree that we'd like to ignore these shifts as well, so I filed:

https://bugs.chromium.org/p/chromium/issues/detail?id=1212783

Once fixed, these shifts should no longer contribute to CLS on your site.

--
You received this message because you are subscribed to the Google Groups "web-vitals-feedback" group.
To unsubscribe from this group and stop receiving emails from it, send an email to web-vitals-feed...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/web-vitals-feedback/d9454fbb-2f8e-4807-b0e2-2af215fd3421n%40googlegroups.com.

[Annie Sullivan]

I should also add that the update to CLS has not been launched in Search Console yet, so that is likely unrelated to the layout shifts you're seeing.

To view this discussion on the web visit https://groups.google.com/d/msgid/web-vitals-feedback/CAChM%3DvpLmvt%2B4x1bX3s_fhmiMZfcOk_zu3TamGu6ALFksgP8xg%40mail.gmail.com.

[Brad Billman]

The search console specifically notes that on April 12th there was an update that might change your results.  And the following 5 days we got from over 90% good urls to 0.

[Annie Sullivan]

Ah, sorry, I thought you meant these changes, which are not live yet. It looks like you're referring to the April notification in search console, which is about the changes in Chrome 90. I know the change log is a little hard to read, but all of these changes would result in lower layout shift scores/more pages passing.

Both the upcoming change and a change to ignore back/forward navigations as having recent input would improve your site's scores, but we don't know of any changes to the metric that would have made the site worse in April.

To view this discussion on the web visit https://groups.google.com/d/msgid/web-vitals-feedback/140f7eec-35ab-4a55-9510-17124a66e1can%40googlegroups.com.

[Brad Billman]

It has caused significant panic for june.

[Brad Billman]

The link that search console points to suggests the changes are in effect.  If this is is not accurate, are there other things that changed on april12th/13th that could tank scores so completely?

[Annie Sullivan]

On Tue, May 25, 2021 at 12:33 PM 'Brad Billman' via web-vitals-feedback <web-vital...@googlegroups.com> wrote:

The link that search console points to suggests the changes are in effect.  If this is is not accurate, are there other things that changed on april12th/13th that could tank scores so completely?

This change in score does not appear to be caused by a change in Chrome or the metric. It's close to the M90 rollout, but the rollout didn't really start until several days after.

 

To view this discussion on the web visit https://groups.google.com/d/msgid/web-vitals-feedback/d538b34b-46f2-4c96-bd11-cdaa61a130fdn%40googlegroups.com.

[Omri Ariav]

Hi - do we know when the new CLS measurement will go into effect? Will it be around Mid-June as communicated here? 

[Philip Walton]

Omri, I'm not 100% sure what you mean by "go into effect", but as mentioned in the post you linked to, the Search ranking changes that are rolling out in mid-June will be based off the updated definition of CLS.

I expect Search Console to switch to showing results based on this updated definition of CLS in early June, as will PageSpeed Insights and Lighthouse.

[Philip Walton]

Brad, as Annie mentioned, it does not seem like your change in scores are a result of a change in Chrome or Search Console. All of the recent changes we've made should only improve your scores.

> are there other things that changed on april12th/13th that could tank scores so completely?

Can you check to see if you made changes to your site that were deployed at about the same time? You mentioned that most of the CLS issues were caused by route changes in your SPA, was any code that could affect that part of your SPA made around that time?

Also, can you share the URL of your site? (Even privately if you prefer not to post it here.) Without more information we can't really determine what could be causing the issue you're seeing.

Lastly, as I mentioned above, the CLS that's happening as a result of users clicking the back/forward buttons in an SPA is something we plan to fix in a future Chrome version.

[Brad Billman]

From everything I can tell.  There were no site wide updates that could have caused such a drastic shift.  We went from 250k-340k impressions on 94% good urls.  To 0% of impressions on good urls from Apr 13 -17.  Web console shows on the day before the cliff that they made changes to CLS.  I've spent hundreds of hours trying to figure it out.  I can see with the variety of tools that route changes seem to have the biggest impact on cls.  Which it sounds like the newest update is supposed to help.  But I don't understand how we were 90%+ good previously, nothing changed with the routing.

The tank did get me deep into web vitals and heavy analysis of the site.  I did find a couple of places that we were able to make positive UX improvements.  But all of that (literally 100s of dev hours) have only gotten us to 100% urls are 'needs improvement'.  Better than poor, but I'm at a loss at this point, and can only hope that the next CLS update that is supposed to be more fair for routing works.

Thanks for your response.

[Rick Viscomi]

Search Console groups URLs together and assesses the entire group's CWV performance. So for all URLs to go from "good" to "needs improvement" all at once, that suggests two things:

1. All or most of your URLs were placed into a single group

2. The aggregate CWV experience for one or more metrics were very close to the "good" threshold, so a small regression in performance or a change in relative page popularity tipped the group-level aggregation over the edge to "needs improvement"

The good news is that it should be just as easy to tip the URLs back to "good" as it was to "needs improvement" assuming that the aggregation is now similarly just beyond the threshold.

The next step I'd recommend is to look at your most popular pages and assess the biggest opportunities for improvement. For example, if you have a very popular page with "needs improvement" CLS, optimizing that page may be enough to get the entire group back in the green. That said, it's a precarious position to be in, so continue to focus on improving your site's performance to avoid being so close to the threshold.

Disclaimer: I haven't seen any data for this site, I'm just guessing what's going on based on the description of the problem.

Rick