cloudflare base url?

Cloudflare is a different kind of CDN-solution Quinn, the CDN-fiels in AO does not apply in that context.

Hope this clarifies,
frank

I downloaded this plugin and also had the same query but when I was configuring CloudFlare I noticed they also offer Javascript, CSS and HTML minifying and Rocket Loader to help run JavaScripts files faster. With all due respect does this then make this plugin redundant for websites using CloudFlare with one less plugin to run?

hi fernuw;
the main difference is that AO aggregates all CSS/JS and then minifies it, whereas CF “only” minifies and applies the rocketscript magic if configured to do so.

but I would advise you to simply test the different scenario’s (ao-only, cf-only and ao+cf) and go with what gives you (with your specific context of theme & plugins) the best results ??

hope this helps,
frank

Hi Fernus,

The minification feature offered by CloudFlare is conservative (to say the least) and nowhere near as effective as Autoptimize. Further, CloudFlare’s feature will not, because it cannot, combine your CSS or .JS files*

*Rocket Loader will combine your .JS files but is ultimately a poor solution as the concatenation is done ‘on the fly’; and, the asynchronous loading method employed by Rocket Loader is non-standard, therefore working in a far more limited number of browsers/devices/scenarios than does Autoptimize’s standard ‘defer’ and/or ‘async’ solution.

Lastly, Rocket Loader is highly likely to significantly slow your site’s page-to-page navigation. If you need one last reason, Google does not recognize Rocket Loader’s proprietary “RocketScript” attribute (even though Rocket Loader does ‘work’ in Chrome).

TL;DR: Autoptimize FTW ??

Best,
AJ

Thanks for the replies Frank and AJ.

Having read various posts and blogs about not running two or more plugins together which cover similar objectives I was initially worried about the consequences of experimenting but a few days ago I read Frank’s responses to reviews left previously encouraging trial and error of the settings available.

Over the last couple of days I am finding for our website basic settings work best and PageSpeed scores for the home page improved from 43 mobile and 63 desktop to 48 and 71 respectively this is also with the assistance of Wordfence’s falcon engine. The performance difference is even greater for many of the other pages.

Thanks for explaining why Autoptimize is superior to similar features offered by CloudFlare. I was lead to believe up until now the fewer plugins the better, I guess this is an exception to the rule! To be honest, I’ve not bothered comparing benchmarks between this plugin and CloudFlare because the latter have pee’d me off with the setup issues I experienced and there lack of reply when I needed help.

Do bear in mind, Fenrus, that PageSpeed Insights is not actually a speed testing tool (you’re not given any metrics), but a ‘Best Practices’ tool; and the implementation of Best Practices may or may not = a faster website. It’s just not that simple. To illustrate this point, some of the fastest website’s on the Internet score terribly on PSI.

Should you want to do real speed testing (I think I can speak for Frank here, too) you will want to use WebPageTest.org (also primarily developed and supported by Google).

Best,
AJ

Hi AJ,

That is true but if PageSpeed is a tool Google is using when considering where to rank me for SEO purposes then it makes sense to perform well in it.

Hi Fenrus,

That Google uses their PageSpeed tool to rank websites is an unfortunately popular myth that Google does a poor job of dispelling. Said differently, there is no causative relationship between one’s PSI grades and SERPs rankings.

Google gathers Crowd Sourcing information from users of Chrome to aggregate real load metrics for websites, with their primary concern being the speed with which above-the-fold content renders and can be interacted with (ergo, all the PSI harping about render-blocking content).

As such, the metrics that do correlate with SERPs rankings are TTFB (Time to First Byte), Start Render Time, Time to Above-the-Fold Visual Completion and TTI (Time to Interactivity).

These metrics can be accurately garnered with WebPageTest.org. The lower you can get your site’s Speed Index, the better. ??

Best,
AJ

Hi AJ,

Thanks for explaining that, I suspect my original perception is perhaps shared by many others so hopefully others will read this thread and help to educate them too.

That Web Page Performance Test is really good by the way! Does a damn sight better job than PageSpeed to show where the website is lacking not that I’m entirely sure how to fix the problems but hey.

Having picked 3 pages at random with grades of DAAFD, FAAFD and DAAFF shows just how much more tuning is required. ??

What I find confusing is that I have updated the .htaccess file to take of leverage browser caching and yet this stubbornly continues to be suggested by various benchmarks.

Thanks again for your sharing your knowledge.

AJ: I think I can speak for Frank here, too

If anyone can, you can AJ ??

AJ: Google gathers Crowd Sourcing information from users of Chrome to aggregate real load metrics for websites

Didn’t know that, thx!

Fernus: Having picked 3 pages at random with grades of DAAFD, FAAFD and DAAFF shows just how much more tuning is required. ??

The first letter indicates your time to first byte is high. If this is for pages that weren’t in Wordfence’s Falcon cache yet, then that’s understandable, but if this persist after the first load (I usually have webpagetest do 5 runs and focus on the median result) then you might have to look into Falcon’s settings (or consider an alternative).

Fernus: What I find confusing is that I have updated the .htaccess file to take of leverage browser caching and yet this stubbornly continues to be suggested by various benchmarks.

If this is the case for Autoptimized CSS/JS, you could switch to dynamic files to deliver the autoptimized files (last option on settings page), this will help take care of compression & expiry of those files.

have a nice weekend,
frank

Couple more things, Fenrus, and I’ll stop sticking my big nose in Frank’s support ??

1.) It’s not that you shouldn’t bother with trying to get your site’s PSI grades as high as possible (which is another way of saying that one should indeed attempt to implement all possible Best Practices). It’s that the implementation of Best Practices, insofar as that correlates with high PSI grades, does not always bear a relation to a site’s real performance metrics.

2.) Crowd Sourcing is in all likelihood not the only means by which Google garners real world performance metrics; it is entirely possible that they also use aggregated Google Analytics data. Further, Google may also employ algorithms similar to those used in the PSI tool.

The point, however, is that Google must needs some or another way to actually get real performance data ‘from the field’, as it were. From the real world. And, since Google are at least as smart as anyone that enters our orbits, we can know that Google knows that PSI grades (i.e. the implementation of Best Practices) does not provide an accurate, real world measurement of a site’s real world performance. As such, the superior tool is one that best replicates real world speed metrics, as that then replicates what Google uses and must use. WebPageTest.org is the by-far-and-away superior tool, therefore.

Luckily, we all have Autoptimize to assist us in making our sites as performant in the real world as they can possibly be. ??

Best,
AJ

Hi Frank and AJ,

Apologies for the delay in responding, I’ve since not managed as much work as I’d have liked but there have been a couple of bereavements over the last few days.

Frank: The first letter indicates your time to first byte is high. If this is for pages that weren’t in Wordfence’s Falcon cache yet, then that’s understandable, but if this persist after the first load (I usually have webpagetest do 5 runs and focus on the median result) then you might have to look into Falcon’s settings (or consider an alternative).

As per your suggestions Frank, I have attempted to address the first grade of the Web Page Performance Test, sometimes a D and other times an F with the Wordfence Falcon Engine. As settings are very limited with this caching engine I deactivated that part of it and instead tried experimenting with WP Fastest Cache which really only increased Yslow scores in gtmetrix by a couple of points but returned consistent F scores for the first letter in Web Page Performance Test. I will try one more caching engine tomorrow but I suspect tangible improvements to loading times and benchmark scores are going to come from lossy compression of images in carousels. There are other uncompressed png images used but I don’t know how to access them and they are not stored in the media library.

Frank: If this is the case for Autoptimized CSS/JS, you could switch to dynamic files to deliver the autoptimized files (last option on settings page), this will help take care of compression & expiry of those files.

I tried this unfortunately the leverage browser caching issues remain. However, I must say I have greater respect for your Autoptimizer plugin. Fastest Cache have their own minification tools and with them activated and your plugin not in use it couldn’t touch the performance achieved by yours in GTmetrix and PageSpeed.

AJ, I understand what you’re saying so now use PageSpeed, GTmetrix and Web Page Performance Test.

I will try one more caching engine tomorrow but I suspect tangible improvements to loading times and benchmark scores are going to come from lossy compression of images in carousels.

image compression surely will help, but won’t improve the base page TTFB. once your page is in cache, your site should fly, really. is the TTFB of the other requests also high? do you see the same issue when selecting another test-location (one near to where the serer is)?

I tried this unfortunately the leverage browser caching issues remain.

even for the autoptimized .php-files? that’s very weird.

frank

image compression surely will help, but won’t improve the base page TTFB. once your page is in cache, your site should fly, really. is the TTFB of the other requests also high? do you see the same issue when selecting another test-location (one near to where the serer is)?

True re image compression, but I’m out of ideas how to improve TTFB. Wordfence’s Falcon engine occasionally lifts the grade to a D whereas Fastest Cache drops it to a consistent F but increases the Yslow score in GTmetrix by a point or two over Falcon. The only other thing I can think of to aid the issue is to upgrade the hosting package but I can’t really afford that right now.

Don’t know if this sounds daft but once you’ve activated a caching engine how to do you ensure the page is in cache? TTFB of all pages that I’ve tested on the site seems to be lowly D’s or F’s if that’s what you’re asking? When you question whether I see the same issue when selecting another test-location do you mean if I see the same scores if I run the benchmarks from another PC? In terms of being nearer to the server that’s apparently based in the Netherlands!

even for the autoptimized .php-files? that’s very weird.

I’m not sure, I’m not knowledgeable enough to answer that yet! What I do know is PageSpeed keeps telling me and so does WebPagetest with its D (69%) score. PNG files mainly in the header seem to be the culprit for the lacklustre leverage browser caching score if that helps?

Shane

hmm, could you enable Wordfence Falcon, configure AO to delivery the optimized files dynamically (i.e. not statically) and provide me with the URL of your site? I’ll have a quick look myself then.

frank