CrUX API: querying with a regexp and a time frame

[Anthony RICAUD]

Hello,

The CrUX API is a valuable tool to make sure the changes you're making are going in the right direction. I have two feature requests that I think will make it even more useful, if they can be implemented.

1. Allow querying a group of URLs with a regexp
   When you're trying to improve a page type, the data for a single page is very volatile. Unless you've made a sizeable change, there's little chance to deduce anything from historical values. By using a regexp that matches say a server router, we could see if there is a trend on that page type. This would also help with monitoring data over a long period: for an e-commerce site, a particular product details page will eventually receive no or very little traffic so you need to change it regularly.
   This could also help websites with less visits to get more data points.
   
2. Asking for data over a smaller window than 28 days
   When you ship a change, only a few days later (5 to 10 in our experience) will you start noticing a trend. If we could query the last 7 or 14 days, this would reduce that lagging effect and make it easier to investigate (if it's a negative trend).

Obivously, those are things you can do with RUM tools but it's nice to get confirmation with a dataset that is closer to what will be used in the page experience update.

Thanks,

Anthony.

[Barry Pollard]

Ohhh massive +1 for ask 2!

[Ryan Siddle]

Hi Anthony,

 

It sounds like you may be better off rolling out your own real-time data collection using the Web Vitals JS (https://github.com/GoogleChrome/web-vitals). I’m doubtful that Regex would be considered for the API. It’s a heavy operation for such dataset.

 

By rolling your own solution rather than using the CrUX API, it is possible to see the events as they occur, along with isolating spit testing for rapid feedback. Aggregating the page types would provide you with a much more usable dataset as you mention. How many events are you expecting per day?

 

Best regards,

 

Ryan

 

Ryan Siddle / Digital Director
ryan....@merj.com

Merj
+44 (0) 2033 222 660 / Ext: #101
9th Floor, 1 Ropemaker Street, London, EC2Y 9HT
merj.com

--
You received this message because you are subscribed to the Google Groups "Chrome UX Report (Discussions)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chrome-ux-repo...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/chrome-ux-report/b6aa8fff-9e3c-4269-b759-325d201f1156n%40chromium.org.

[Anthony RICAUD]

Hi Ryan,

My clients are already using a RUM solution (or it's the first thing we set up). Even though this is the best tool to guide a project, I always want to check the impact in the data source that matters: CrUX. Hence my requests. I don't see that as a replacement for RUM but a complement.

[Rick Viscomi]

Hi Anthony,

Thanks for the feature requests! I'll pass those along to the engineering team.

Are you also using the CWV report in Search Console? That tool groups similar pages together to show aggregate CWV performance, along the lines of your suggestion.

I'd also echo what Ryan said. I know you're already looking into RUM but I'd emphasize that it should be your first line of defense for monitoring CWV. The public CrUX dataset will eventually catch up, but it's not designed for immediate feedback.

Rick

[Anuj Puri]

Hi Rick,

When you say " The public CrUX dataset will eventually catch up, but it's not designed for immediate feedback."

-What is the specific time that it takes for the results to show up in CRuX

-Also, does this mean that unless the data reflects in CRuX, the improvements made will not be accounted in search rankings?

Thanks

Anuj Puri

[Rick Viscomi]

Hi Anuj,

What is the specific time that it takes for the results to show up in CRuX

There wouldn't be a specific day that the change suddenly appears, but rather it would gradually move the distribution, depending on factors like the magnitude of the change, affected users, etc.

Also, does this mean that unless the data reflects in CRuX, the improvements made will not be accounted in search rankings?

Ranking questions are best addressed on the Search Central forum.

Rick

[Anuj Puri]

Got it. Thanks Rick!

[Anthony RICAUD]

Thanks for passing this along.

I also use Search Console tools and it's useful to discover which types of pages to focus on. Besides that, it's less useful for day to day work because it's difficult to extract the data (no API, the export button only exports the number of URLs in each bucket). The grouping is most of the time useful but I often notice a kind of "miscellaneous" group for page types that only have a handful of URLs (like landing pages, about pages, etc).

Instead of a regexp in the CrUX API, maybe we could use structured data on the page? If we declare a page type, both the CrUX API and Search Console could use that hint to generate groups (and I say hint because there are many reasons for Google tools to group in another way).

About the time frame request, let me highlight why RUM is not enough (but totally agree that RUM is a critical aspect of monitoring CWV):

• RUM can't account for iframes because the browser APIs have privacy limitations. An ad banner could be the LCP in CrUX but not in RUM. Some third party content could generate a lot of layout shifts in CrUX but be invisible in RUM.
  
• The population captured by RUM and CrUX are different and there's no way to filter the RUM data to match the CrUX population.

With those differences, any CWV effort to get the upcoming search ranking benefits needs to pay close attention to CrUX data even with a solid RUM in place. With an additional shorter time frame, I believe this will help those efforts.