Page speed - what do you aim for?

[McTaggart]

Hi Mozzers - was just looking at website speed and know the google guidelines on average page load time but I'm not sure whether Google issues guidelines on any of the other 4? Do you know of any guidance on domain lookup, server response, server connection or page download?

Page Load Time (sec) - I tend to aim for 2 seconds max: http://www.hobo-web.co.uk/your-website-design-should-load-in-4-seconds/ 
Server Response Time: [Google recommends 200ms]: https://developers.google.com/speed/docs/insights/Server Redirection Time (sec) [dependent on number of redirects so probably no guide figure]
Domain Lookup Time (sec)
Server Connection Time (sec)
Page Download Time (sec)

Thanks, Luke

[lcallander]

IMHO, if somebody is paying us for SEO, then our GOAL is to get the homepage to load in a second or less.... especially if most of the users are mobile.  If it's mid 1 second, then we can grudgingly live with that.

I'm glad you asked about server response times.... for most sites, after the content is optimized ( smaller images, clunky code, etc...) the initial server response time is usually the culprit for getting over a second.... as long as the rest of the home page is "light".  Light to us is under 1MB.  Depending on your CMS, there are a variety of ways to get the response time to be 200ms or less.

Google Pagespeed, as David said, is a good measurement, but it's not the holy grail of measurements.  We use it only to identify areas that need improvement.  Waterfalls tell us what's taking so long and what's heavy.

You didn't ask about plugins - which is a major culprit to caching, minify errors, conflicts, speed and weight.  We limit all active plugins to TEN (including caching, SEO, security).  For some sites, plugin clean up is the easiest way to speed up a site.

At the end of the day, nothing beats clean code, light images and a lightening fast server.

[McTaggart]

Thanks for all the feedback everyone - much appreciated, Luke

[David-Kley]

As long as the page loads quick for users then I wouldn't put a huge focus on this. True that Google looks at page load speed, but I wouldn't put all your eggs in that basket. We have sites that show a 2.5-3.5 second load time, and they still dominate ranking results. Focus on creating a better experience.

One of the simple ways to speed up load times is to minify and compress CSS and Javascript files as small as possible, but be sure to check that the minification does not break areas of the site. We have seen improvements as high as 75% just from completing this step alone.

If you have a Joomla or WordPress website, here is a great plugin that will do this for you: https://www.jch-optimize.net/

[BlueCorona]

Hi Luke! When using this tool (https://developers.google.com/speed/pagespeed/insights/) we aim to have our clients above 80 for both mobile and desktop.

[LesleyPaone]

I will be honest, I don't trust Google with PageSpeed. There are too few questions asked about how it actually fits in with the metrics and what is used. One example is Google says resources like Google Analytics do not count against your score in the SERPS. But in the test they do. If you use several Google assets like Adwords, Analytics, fonts, ect; you will show a very low score. Using them will actually block you from seeing other things that you can fix.

What we have started doing is figuring out what Google actually needs and presenting it to them. We started hiding tracking codes from Google. Bing, Facebook, ect from them. We hide our analytics tracking script from Google's crawler. I figured out that Google's test servers have the FA library and also their font library locally on the machine, so we have started hiding those from it as well. Any 3rd party script we have that Google does not need to see has been hidden as well, segmentify, olark, anything really. Doing these things has raised our score quite a bit.

[ThomasHarvey]

"if you have 2 or even 3 redirects mobile users wait for 5 seconds before see anything. Hint - that's why i won't click on most bit.ly, ow.ly, goo.gl links in Twitter, Facebook, G+ when i'm on mobile. Because they first pass via t.co redirect then redirect that i can see and sometime even 3rd redirect."

Just adding a bit of weight to what you said, here's a test of a t.co link through bit.ly:  https://i.gyazo.com/ca87c486a903914c2b058612cc93f3f0.png on 3g, it's 4.27s to even start loading Google. Without t.co: https://i.gyazo.com/f22c18a0879f76ecf653662153e17c43.png which is 2.35s.

[ThomasHarvey]

Pagespeed score means nothing unfortunately. http/2 puts a spanner in the works for a lot of it.

https://blog.newrelic.com/2016/02/09/http2-best-practices-web-performance/

Being this section:

• Concatenating JavaScript and CSS files: Combining smaller files into a larger file to reduce the total number of requests.
• **Image spriting: **Combining multiple small images into one larger image.
• Domain sharding: Spreading requests for static assets across several domains to increase the total number of open TCP connections allowed by the browser.
• Inlining assets: Bundling assets with the HTML document source, including base-64 encoding images or writing JavaScript code directly inside

[Mobilio]

It's hard to be explained but "Less is MORE!" in general for that numbers.

Examples - redirectors. Redirects can overkill your site specially on mobile users. For that even simply site redirect can took second or two. Example www.example.com -> 301 -> m.example.com; looks simple isn't? But in reality after client took 301 redirect he must make new domain resolving (for m.exmaple.com) and then new connect to new server (m.example.com). And this is simply case... if you have 2 or even 3 redirects mobile users wait for 5 seconds before see anything. Hint - that's why i won't click on most bit.ly, ow.ly, goo.gl links in Twitter, Facebook, G+ when i'm on mobile. Because they first pass via t.co redirect then redirect that i can see and sometime even 3rd redirect. I know that marketers want to see "clicks", but isn't good for mobile users.

Server connection is also need to be less. But this mean that server need to be closer to user. Best example is Australia. There even simply DNS resolving + connection took one second. And client doesn't receive single byte from server yet... You can see WebPageTest.org (there are Australian servers). But of course providing single server there is expensive, so you need to have deep pockets to make servers there. That's why most of companies providing CDN support. Since CDN endpoint is closer to user it make things little bit faster for them. And if CDN is setup correct should be much faster.

So - idea is "Less is More!". The best is if you use WPT to benchmark your site from all over the world. And also setup Analytics to count speed. Because it's different speed when your site is on perfect conditions in datacenter than in real world.

[Vijay-Gaur]

Hi Luke,

Here is what google recommends in terms of page speed. Server response time to be less than 200 ms.

Now, coming to the Page Speed tool / Insight that google provide the measure the page speed ratings (1-100) , Google Page Speed score is indeed a strong indicator of a website’s loading performance in terms of time.

As per my research, total website download less than 10 secs corresponds to 75-85 on pagespeed score.

I hope this helps.

Thanks,

Vijay

[McTaggart]

Thanks Tom for picking up on that error - ugh - corrected now. Brain working sluggishly this morning lol!

[ThomasHarvey]

Hi Luke,

"Avg. Page Load Time (sec) [Google recommends 200ms]:" That's actually for the server response time.

Personally, the only thing that matters is that the overall page load time is quick. I aim if possible for sub 2 seconds for any page.

Tom