How complicated would it be to optimize our current site for the Safari browser?

[danatanseo]

Hi all! Okay, here's the scoop. 33% of our site visitors use Safari. 18% of our visitors are on either an iPad or iPhone. According to Google Analytics, our average page load time for visitors using Safari is 411% higher than our site average of 3.8 second. So yes, average page load time pages loading in Safari is over 20 seconds...totally unacceptable, especially considering the large percentage of traffic using it.

While I understand that there are some parameters beyond our control, it is in our own best interest to try to optimize our site for Safari. We've got to do better than 20 seconds. As you might have guessed, it's also killing conversation rates on visits from that browser. While every other browser posted double-digit improvements in conversion rates over the last several months, the conversion rate for Safari visitors is down 36%...translating into 10's of thousands in lost revenue.

Question for anyone out there gifted in Web design and particular Web Dev....Do you think that it's possible/reasonable to attempt to "fix" our current site, which sits on an ancient platform with ancient code, or is this just not realistic? Would a complete redesign/replatform be the more realistic (and financially sound) way to go?

Any insights, experiences and recommendations would be greatly appreciated. If you're someone interested in spec'-ing out the project and giving us a cost estimate please private message me. Thanks so much!

[danatanseo]

Thank you all so very very much. Matt, I am going to drop you a PM as you suggested. Paul, wow...thank you for sharing your insights here. I am sure this is extremely helpful information not just for me, but for many other folks here who've observed some of the same things on their sites. Yes, I agree that the conversion issue could very well be a mobile optimization problem and not a Safari problem as we have spent very little to no time optimizing our mobile site. Given the traffic levels coming in via that channel, it's probably time to get crackin'!

You guys are awesome!

[ThompsonPaul]

Sorry Dana - totally missed your mention about conversions being down. In that case you'll want to do some head-to-head testing with webpagetest.org.

The other thing to consider is that over half your Safari traffic is from mobile, Is it possible there's a mobile-optimisation problem and not a Safari problem? (How do the other mobile browsers perform in comparison?)

And the last suggestion I just thought of: Pingdom now offers Real User Monitoring (RUM) that purports to also be able to track the page speed of Safari users. (It uses it's own javascript module you add to your pages.) Given that this comes free for 1 site even with their free plan, seems like it would be an essential step toward getting a better handle on just where and how big the issue actually is.

P.

[ThompsonPaul]

Dana, the good news here is that you have probably been worrying for nothing.

The function GA uses for tracking page timing is simply not included in Safari or Safari for iOS (phone/iPads). So those Safari figures you are seeing are completely anomalous and totally unreliable.

Here's the chart that shows which browsers have the capability to report their page speed in Analytics.
http://caniuse.com/nav-timing
As you'll notice, both desktop and mobile versions of Safari are no-go.

The bad news is, just because the GA timings are worthless doesn't prove definitively you don't have any Safari issues. By far the best way to assess the situation is to analyse whether Safari-based visitors behave significantly differently on-site. Is their bounce rate higher? Time on site or conversions substantially lower? If no substantial difference compared to other browsers, you don't have an issue. (Usually this is the case - a site that performs in Chrome and Firefox almost always performs about the same in Safari, in my experience)

The one option you do have to get actual speed tests is to use webpagetest.org and select a testing location that offers the Safari browser as one of the options. Then you'll want to test multiple pages that represent the different primary templates and sections of your site. Do note though that even this will be using Safari on Windows, which was capped at version 5 back in 2012. So you can expect that current versions of Safari, especially native Mac apps, are likely faster.

Hope that sets your mind at ease, and gives you some next steps to further investigate alternative methods!

Paul

P.S When using GA page speed results, do note that by default Analytics only gathers data on 1% of page views. You need to customise your GA site snippet to push that level higher. This is critical for most sites because unless they have huge traffic volume, a 1% sample rate will be wildly inaccurate. The code snippet to add is

_gaq.push(['_setSiteSpeedSampleRate',100]);

and it has to go before the call to _trackPageview. By setting the sample rate to 100 as above, GA will try to track speed of all pageviews. It won't succeed of course, but you'll have far more reliable data. (GA will record speed data for a max of 10K hits per day)

[Matt-Williamson]

Hi Dana,

I have done lots of work on site speed (a little obsessed with it but I have seen great results!) and I have made changes to a lot of sites that have massively improved load time and I have checked a couple and they are loading as quickly in Safari as the other browsers. I would be happy to give you some advice on possible quick fixes that you could make relatively easily and they will speed your site up significantly if you haven't already implemented them.

Drop me a pm with your site address and what server setup you are on Apache or IIS. I will have a look and see if there is any changes that may help

[Matthew_Edgar]

Hey Dana,

I've had this problem with Safari being slower as well. In fact, I just checked a handful of sites in Google Analytics and Safari is almost consistently the slowest browser. One question to look into is whether or not mobile traffic on Safari is slow. For the sites I'm looking at and for the sites I've worked on in the past, the culprit is usually Safari's desktop browser. Phones are usually within normal loading parameters (by phone standards).

Unfortunately, in most cases, the sluggish performance on Safari is due to Safari's DNS prefetching. Generally, prefetching does the opposite, but apparently can slow you down in Safari. You can read more about that here. http://macs.about.com/od/MacTroubleshootingTips/qt/Troubleshooting-Safari-Slow-Page-Loads-Caused-By-Dns-Prefetching.htm and this is another good one http://computers.tutsplus.com/tutorials/how-to-fix-slow-and-non-loading-webpages-in-safari--mac-51338 (Lots more if you Google Safari prefetching.)

I do think it is worth it to try to fix it. Here's what I would do, if I were you. Feel free to PM me if you want more help...

The way you fix this is on the client's machine, at least according to those articles (and others I've found while Google'ing around just now). However, you can try these steps out on your computer (if you've got a Mac) and see if you have an improved performance in Safari. It shouldn't take that long to test, and is probably worth it given the huge loss you are seeing due to this issue.

The next question is, well, how do I do this for everybody? The answer is to ask your visitors to use Chrome! No, seriously, one that worked for me in the past, was to detect the Safari user agent and load a slightly different version of the page that required fewer requests, meaning there will be less for Safari to prefetch. For instance, you might get rid of JavaScript, images, etc. that isn't essential. Obviously, keep the core content so that the pages are basically the same (kind of like you would for mobile detection).

This is obviously much more complicated to setup as it requires adjustments to the design as well as the code structure. But, generally simpler than trying to redo your entire website.

Before you make any such changes, it might be worth running a simple test on a key landing page that gets a lot of visits from Safari. Remove what elements you can for all user agents and see if this changes anything substantially with the site speed in Safari.

I hope that helps. Thanks,
Matthew