Check your pageload speed:If it takes more than one second to load, make changes.

[Beachboy]

I haven't posted here in a long time, but I slammed from page 7 to page 2 in Google SERPs in a week by improving load speed. Using a website page load test site whose name begins with P, load time improved from 3.4 seconds to 800 milliseconds (8/10th of a second) by:

1. Optimizing images to the minimum file size while still proving a quality image.
2. Changing to a lightweight image slider.
3. After doing online research, replacing bloated plugins with lighter weight versions.
4. Any plugin not absolutely needed, in the opinion of a web consultant, deactivated or removed.
5. Changed from shared hosting to a virtual private server hosted in the "cloud."
6. Began using a CDN (content delivery network).

Next step: Looking into a lighter weight WP theme and page building system.

Many of those who still rank above me are about to be surpassed. Most of them have page load speeds slower than mine. Check six, guys.

[Wilburforce]

My pages have loaded very fast (<700ms) in the UK for some years. Google probably looks at speed from US servers, which will inevitably cause some delay (although they nevertheless all score 100% on Google's test), but my target market is UK locality, so apart from improving the US load time for Google there would be no benefit in using a CDN.

All the measures you suggest can help, as can looking at all the components in Google's (or other) speed test analysis. However, all the tests I have seen analyse one page at a time, and it is also important to address things that affect real-world speed for users, like ensuring all your plug-ins - and anything else that features on multiple pages - is cacheable: most of us hope that users will visit more than one page, so (e.g.) your logo needs to be in a form and format that avoids the need to load it with every new page.

As you may gather, I fully agree that page speed is important - mobile users with poor reception are not going to wait indefinitely - but as far as Google SERPs are concerned I think its effect is marginal. Other factors carry much more weight.

[robzilla]

Google probably looks at speed from US servers

The speed data is sourced from the Chrome User Experience Report [developers.google.com], so from your Chrome-using visitors (if they've opted in to sharing this data with Google when they installed Chrome). If your users are mostly from the UK, the speed data should reflect that.

[Wilburforce]

@Robzilla - Thanks! That's useful to know.

I usually do speed tests on Firefox, which is probably why the Google test shows slower load speed than tests on which I can select UK servers. Next time I check I'll do an A/B between Firefox and Chrome.

Another factor that makes quite a difference - probably why my pages still test OK from the US - is reducing trips to the server. Consolidating everything you can makes a real difference, so try to avoid multiple CSS and JS files, and where images are for illustration only (i.e. when they are not part of the page narrative), use sprites. Thirty trips to a CDN server can take longer than three trips to a distant one. Also - my earlier post refers - it may take longer to get a big CSS file that contains a lot of irrelevant CSS for the current page, but if it only needs to be fetched once for the whole site then all the remaining pages will load faster.

[iamlost]

Just a note of caution: optimisation is critical, however is quite different for HTTP1.1 or for HTTP2. What works best for one can truly bog the other.

Further: while various speed tests are handy guidelines it is wise to remember the varying lag times in the different connection protocols, that on the same same connection mobile may be twice as slow as desktop, etc.

And that the bottom line reason for fast render is UX and CR. If you have family and friends in other regions, countries ask them to do time tests on a variety of devices in a variety of locations. The results may depress you. Or not :)