Does Page load speed matter?

The short answer is yes – if you care not to frustrate, drive away and maybe lose forever a proportion of your audience.

…oh, and their friends.

I was recently provided with an ad’ banner ( A Flash-based one) to embed into a site homepage, and questioned the 3Mb file size. Embedding this would more than double the download size of the page and no-doubt impact on load time.

The assumption is that most of this site’s visitors enjoy good quality broadband or 3G, so worrying about download overhead is a last-century concern, surely?

Page speed remains a modern concern. Research undertaken between 2009 and 2012 by Akamai, Google, The Aberdeen Group and others shows that there are short and long term effects to the most minor of delays in page load speed.

User response to minor delays can be startling:

  • A 1-second delay in page load time results can result in a user viewing 11% fewer pages, and a 7% drop in conversions (source:
  • Half of web users expect a site to load in 2 seconds or less, and abandon it after three seconds (source:
  • 57% of online consumers will abandon a site after waiting three seconds for a page to load, 80% of these will not return, and half again will tell their friends of their poor experience (source:

Google’s test subjects showed a measurable behavioural change when search results were delayed by as little as 100ms and 400ms. These tiny delays were enough to reduce the number of searches the subjects went on to  perform. Prolonged exposure to a half-a-second lag had a long term effect, with subjects taking weeks after the lag was lifted to recover their usual search behaviour.

Speed kills

For the site in question we cannot (with our current analytics) judge with certainty what effect page speed is having on our visitor behaviour. Google Analytics’ Site Speed feature tells us which pages take the longest to load, and we’re able to see that in periods where overall page speed is slower than usual (for whatever reason) that we see a reduction in pageviews for the duration.

Hopefully the addition of goals and examination of conversion rates will help us track the effect of page speed in isolation from other factors.

…and size matters

Benefitting from an excellent network infrastructure, its rare to experience any lag when loading own-site pages, but Site Speed reported some figures that required examination.

Employing data from Google Toolbar users, Site Speed records real-user load times for pages. Moreover they represent total load time, so might include embedded feeds, social media plugins and those features we load asynchronously a second or two after the page renders.

…but when you load matters more

Examining the larger pages we could see that some of the lazy-loaded content was bulking up our figures. So if we’re already prioritising inline content, lazy-loading secondary items and loading other items on demand, do we have anything to worry about when it comes to page load speed?

As one of 200 measurements used to rank pages at Google, speed is as yet a minor component. As Google’s Matt Cutts outlines here, this factor will most likely affect sites that are identical in content to their rivals – but for the user-experience, cracking page load speed is much more important.

Even more so when you see how web users actually perceive page load delays. Strangeloop’s excellent infographic on the effects of load time demonstrates how users perceive load delays to be worse than they are, and tell friends they were far worse again.

Users first

For me, considering page load speed as it affects the end user makes it a worthwhile endeavour to both periodically check real-user load times of high-profile pages and make overall page size and suitability of content a consideration in new developments.

There are so many possible reasons why a visitor might experience a half-second delay – from peak-time ISP throttling (my personal bugbear), browser add-ons and having too many processes running on your low-end Ipod Touch to the joys of 3g latency, its best not to have unwelcome content, unnecessary script inclusions and under-optimised media make matters worse.

By far the best thing we can do is determine what a user wants from a resource, and make sure we provide it, and make it as easy to take in as possible.

If half-a-second lag is too much for some, what must half-a-second’s initial comprehension do to their assessment of a page’s value, and its likelihood to provide what they’re looking for?

It may not be measurable in this way, but page speed should be considered from the moment of request to the point of user understanding.