The case for sub-one second mobile site performance

Back in August, Google’s Webmaster Central blog posted an update stating its ambition to get mobile pages loaded in under a second, citing research by the Nielson Norman Group showing user flow becoming disrupted at this threshold.

It also referred to analysis of mobile device loading trends showing global mobile load times are dropping year on year, but still average just over seven seconds

So we can see Google’s ambition and reasoning stated pretty clearly. More importantly though, it’s been re-iterated by Matt Cutts over the last few months as well.

Matt was the conduit for explaining and supporting the introduction of page speed as a ranking factor for ‘desktop’ search (AKA, Google’s main algorithm) back in 2010, and he’s performing the same role for mobile today.

Matt talked about penalisation of ‘slow loading’ (aka plus one second) mobile experiences at SMX in June, (there’s a good roundup here) supporting an official post highlighting SEO demotion factors for smartphone device searches.

At that stage, the page speed promotion element was not rolled out, and these announcements were simply alerting webmasters ahead of time that they need to get their mobile house in order to avoid penalisation for poor smartphone page load performance.

Matt stated in that appearance at SMX in June that the rollout was “coming soon”.

He reiterated the point at Pubcon in October, (you can view the whole keynote here) stating that webmasters of mobile sites need to “get ready”.

Also related to mobile, and covered in the keynote were:

  • Clarification of ‘Hummingbird’. This is fundamental to mobile, as that’s where the natural language searches are most prevalent, and detection and algorithmic adjustment in line with page speed is baked into the infrastructure changes associated with Hummingbird, including the mysterious ‘third tier’ index also announced (my $0.02 here: it’s an indexing tier tightly related to speed factors, equivalent to Big Daddy’s ‘Supplemental Index’ in another dimension. A post for another day).

  • Smartphone ranking will now exclude sites running Flash assets if the smartphone won’t support it (read: iPhones). This is significant enough on its own, and has slipped under the radar somewhat.

  • Around 40% of YouTube video views are via mobile devices. Explains why Google is being so aggressive regarding mobile!

So we can say conclusively that sub one second will be a metric and baked into Google’s smartphone rankings: in fact, it’s a good bet it is testing live on datacenters out in the wild already.

Delivering speed for global mobile SEO success

If you’re running SEO across a multinational site, then your challenge for executing fast page load times are more extreme than single territory sites, as Content Delivery Networks (CDNs) come into play and may make many of the ‘easy’ optimisation steps impossible (particularly JS optimisation elements), but let’s walk through the basics first, then look at global expansion and tool options.

We can achieve sub-second rendering of the above-the-fold content on mobile networks by applying the following best practices:

  • Server must render the response (< 200 ms).
  • Number of redirects should be minimized.
  • Number of roundtrips to first render should be minimized.
  • Avoid external blocking JavaScript and CSS in above-the-fold content.
  • Reserve time for browser layout and rendering (200 ms).
  • Optimize JavaScript execution and rendering time.

The important aspect to note here is the focus on ‘above the fold’ render times. This is really where Google’s sub-second target is applied, and this is key to achieving a significant SEO competitive edge for smartphone performance.

We know that Google has been executing pages as a ‘headless browser’ for some time now – since 2008, in fact – to render pages when accounting for general page load times, and a similar approach is used for mobile render times.

PhantomJS is a great way to test Google’s indexing experience for different website implementations during page speed analysis, and will apply for mobile responses too.

If you don’t already have a performance test process or environment, you should consider using PhantomJS in a stack with Selenium to automate your testing and ensure your page render times are as slick as possible before putting a CDN wrapper around it.

There’s a great explanation of this stack’s benefits and setup over here which also gives an overview of why this type of testing is useful.

There are mobile flavours you can plug into Selenium to specifically emulate particular mobile devices (see iOS Driver for example).

When thinking about what to change in the code and code layout, Google’s page speed developer guides are comprehensive and effective and as such should be used to build your starting checklist before moving to the testing stage.

After tweaking to get the best results using your test stack, running benchmarks with Google’s PageSpeed Tools will give you a clear picture of your site mobile performance and pointers on areas of remaining opportunity.

After this stage, you need to consider your CDN options, as they will typically apply their own ‘optimisation’ to your pages as part of their caching and edge serving. So begin dialogue to see your options to over-ride these ‘features’ where they break your performance gains.

I’ve written about assessing CDNs for SEO before, so take a look if this is a new consideration or you’re thinking of evaluating your supplier.

Of course, if you’re not serving a site to a high volume global audience and a CDN would be overkill, then much of Google’s best practice can be overlaid on your stack (assuming you’re not tied to MS servers) using their PageSpeed module for Apache & Nginx.

What’s the impact?

Before Google ramped up the importance of page speed for its main algorithm, uptake amongst webmasters to implement Google’s recommendations was poor.

Indeed many today would argue that uptake is still poor. However great strides have been made by sites in competitive SEO niches to deliver fast sites as they have the dual carrot of preferential rankings and better site conversion rates (basically: more traffic, converting to revenue at a higher rate).

As its audience shifts to mobile and table devices (as has dramatically happened over the last year) Google’s incentive for smartphone performance will have a similarly engaging effect.

So if you’re in a competitive niche for SEO (or even if you’re not), you likely have around 12 months of competitive SEO advantage ahead of you from today if you get your mobile site performance up.

I know what I’ll be working on in that time, so expect your competitors will be too!