Google to use Page Speed as Metric

[Brett_Tabke]

Google will make page speed a factor in mobile search ranking starting in July

Google today announced a significant change in how it ranks websites for mobile searches: it will now take page speed into consideration as one of its signals, the company says. The change, which Google is referring to as the “Speed Update,” will go into effect in July 2018, and will downrank very slow websites under certain conditions.

Though speed will become more of a factor in determining the order of search results, the change is not so drastic as to make it the only factor. There will be times that slow pages still rank highly – like when they have the most relevant content related to the search query at hand, for example.

[techcrunch.com...]
[webmasters.googleblog.com...]

[Wilburforce]

As they have been dialling up the focus on speed for some time, this can't have surprised many here.

As they are obviously talking about speed (as opposed to optimisation), it would be interesting to know how they intend to deal with server locations. Will they be evaluating my GB site's page-load speed from Caifornia? Is it time to buy shares in CDN providers?

When they say it "will only affect pages that deliver the slowest experience to users and will only affect a small percentage of queries", it can only mean that a negative weighting (a penalty) will apply to slow pages. It will probably be a small weighting, however, as "The intent of the search query is still a very strong signal, so a slow page may still rank highly if it has great, relevant content." So your great, relevant high-resolution pictures and videos will probably be OK.

[keyplyr]

Their PageSpeed & ThinkWithGoogle speed testing tools deliver conflicting reports.

If you watch your server logs, a measurable occurence of weblight requests may be an indicator of how Google sees your page speed in real time user performance.

[MayankParmar]

So the AMP enabled sites are whitelisted?

[engine]

This is a significant change. However, although it talks a page speed, if you've got a poor web host you could suffer with this. It seems a very blunt tool, imho.

As Wilburforce says, shares in CDNs could be good to buy.

>So the AMP enabled sites are whitelisted?
I doubt it needs to be as AMP pages should be fast and lightweight, and so should any HTML page.

[keyplyr]

As Wilburforce says, shares in CDNs could be good to buy

Placebo affect IMO. There have been equal reports of slower results with CDNs (and a whole lot of routing issues.)

[brotherhood of LAN]

I wonder if geolocation counts. Take sites A and B and all other things being equal, if they serve a particular location and site A is hosted closer (and faster to their customers) it should deserve preferential treatment.

Probably not, would be my guess. "How much stuff to download" and "how responsive the servers are" are more likely prominent.

[keyplyr]

Sites on servers that use SSD and support HTTP/2 should be fine.

[graeme_p]

I like this. Slow sites are a worse user experience.

I find this odd:

Although speed has been used in ranking for some time, that signal was focused on desktop searches. Today we’re announcing that starting in July 2018, page speed will be a ranking factor for mobile searches.

I would have thought speed is more important on mobile.

[keyplyr]

I would have thought speed is more important on mobile.

They pretty much said this in those announcements about the Mobile-First Index [webmasterworld.com] but now they've set a relative date.

[Wilburforce]

Out of interest, has anyone seen any effect of the speed factor in desktop results?

[keyplyr]

You mean last year before the Mobile-First Index went into effect?

Hasn't page speed always had some level of weight in ranking, at least for the last few years?

[engine]

I'd be interested to know how the cut-off will work, or whether it's a sliding scale.

1second to load is good, or is 3 seconds to load too slow and gets penalised/demoted somehow. Will the faster sites be boosted. So many questions and so few answers.

[Webwork]

Living in a world of instant gratification I wonder how long anyone will linger or how long any experience will be . . gratifying?

[Future]

###Google to use Page Speed as Metric in Mobile Search Ranking###

Its too old, since #web_browsers, Google keeps on re-inventing... (FOR PROFIT ? then thats END of Story)

But does it effects real SPAMMERS ? And unethical teams, still find a way.. to get rid of same.

#Google should understand.... A Difference..

Thats all,
I learned from great people here..

[robzilla]

Data from the Chrome User Experience report (i.e. real-world metrics) was recently folded into PageSpeed Insights. While that may or may not be what Google uses in the search algorithm, it might give you a good indication of how your pages are performing, relative to what they consider to be "normal", "average" or "slow".

FCP means First Contentful Paint (visual response), DCL means DOM Content Loaded.

[mcneely]

I would have thought speed is more important on mobile.

Ditto -- being since phones, as a rule, have the tendency to put more of a burden on a network than any other connected device does (50-150 MS on average)-- Having a big processor in a phone only serves to generate a faster speed locally .. Latency issues will most likely be the determining factor here. If you are switched through to a, or have a crappy network to begin with, then speeds may vary. I'm guessing that Google will only be testing speeds as they relate to their network, because using a phone to benchmark speeds through any other number of networks that are out there would be tantamount to trying to fill a bucket without a bottom. Connection speeds on phones are wildly inconsistent and are a poor measure of how well or not a site might perform.

This whole business of using phone connectivity speeds as they might relate to site loading is just another huff-n-fluff piece from Google to keep us chasing our tails - Big words like 'ranking factors' and 'mobile index listings' are used to keep us on the edges of our seats. At the end of the day, it won't really even make a difference when you consider all of the hopping and pinging that phones are required to do just to stay functional on the internet.

[Wilburforce]

it won't really even make a difference when you consider all of the hopping and pinging that phones are required to do

The more hopping and pinging they are required to do, the more difference extra trips to the server will make.

[MayankParmar]

Shares in CDNs could be good to buy.

Cloudflare is not enough. I'm planning to buy Stackpath, or shall I go with KeyCDN? :/ It's really confusing. I'm not targeting any country.

[whatson]

how is speed measured?

[jetteroheller]

I hope they do it. My page is very good optimized for speed.
Have 99 from 100 at pages without AdSense
0 additional round trips, 0,1 MB to render
pages with AdSense
0 additional round trips, 0,3 MB to render

So I hope for great improvements.

[robzilla]

So I hope for great improvements.

You can hope, just don't expect ;-) This is likely to only affect (i.e. downgrade) really slow pages.

[super70s]

I hope they do it. My page is very good optimized for speed.
Have 99 from 100 at pages without AdSense

My main page has no Adsense, is hand-coded, loads really fast and only has an 81 PageSpeed score. I wouldn't doubt if some sites that take 5 times as long to load have a score of 71. I've always wondered about the reliability of these scores.

[robzilla]

You're just being scored against a list of best practices, and the final optimization score has no direct relation to page speed (it's not a speed score). A page with a score of 40 can be faster than one with a score of 99, so this is obviously not what they'd use in ranking. Fortunately they've recently added real-world data to Insights so you can get a better idea of how your page is performing for mobile and/or desktop users.

[Sgt_Kickaxe]

Now on my to-do list is to block all popular speed metrics recording websites to stop them from visiting my pages. You know that whoever holds #1 for a term will get the honor of being pinged to death by the slower site owners because they want to see how they measure up.

I wouldn't put money into CDNs just yet, I'd put some money into the leading speed metric recording sites.

[whatson]

OK how is speed measured? Is it from Google's score of items at PageSpeed Insights? As they do NOT accurately determine a site's speed. Efficient coding and efficient sql queries is what makes a site fast.

[keyplyr]

OK how is speed measured?

I think that's been answered above.

You're just being scored against a list of best practices, and the final optimization score has no direct relation to page speed (it's not a speed score).

[robzilla]

OK how is speed measured?

We don't know for sure, but data from Chrome users seems the most obvious and reliable candidate. This obviously scales much better than testing each page on the Web with a headless browser, and automatically prioritizes performance of popular pages (lots of data) over impopular ones (less data or none at all). Another option would perhaps be to make an educated (calculated) guess based on data from Googlebot, like the number of images, scripts and other resources and their size, the time-to-first-byte (TTFB), etc.

Efficient coding and efficient sql queries is what makes a site fast.

Efficient coding can help a great deal by reducing the TTFB, but there are other things like latency, DNS, page contents and third-party resources that can impact performance negatively. If you want to be fast, you need to cover all your bases.

[keyplyr]

One thing those Google speed tools miss is what a real browser would cache and not need to download again.

[Wilburforce]

One thing those Google speed tools miss is what a real browser would cache and not need to download again.

I would expect any attempt to gauge speed to measure the time it takes for above-the-fold content - including all necessary resources - to render. It couldn't realistically take cached resources into account, although I suppose it could make an adjustment to notional speed based on whether browser caching was set. However, that still leaves the question of whether - where css loading is deferred - default html formats would count as successful rendering.

How big a can of worms do you need to get 100% of your RDA?