Mobile video is a major up-and-coming trend in content, with brands everywhere converging on the new and lucrative mobile video market.
Mark Zuckerberg said on a recent shareholder conference call that he sees video as “a megatrend on the same order as mobile” – which makes mobile video, the intersection between the two, the ultimate sweet spot of engaging content to draw in new consumer eyeballs.
But sadly, there are still some technical hurdles to overcome before the mobile video experience is as smooth as companies would like it to be. In our previous installment we looked at how video can be a massive mobile data hog, and why it shouldn’t (but still does) have an impact on download speed.
In this part we’ll look at the contentious subject of autoplaying videos and their impact on mobile webpage performance, as well as how audio can delay page speed, and what kind of conditions make for a poor viewer experience (VX).
Our third and final part will consider some solutions that webmasters can enact to counter the issues with mobile video.
Video autoplay and page performance
Comparing the data on HTTP Archive for average content for the top 100 most popular sites (according to Alexa) with the top 1 million (shown above) reveals some interesting stats.
On average, video content is just 17kB (rather than 128kB) which is 2.1% of total page size, which, is a (comparatively) slender 828kB.
There are three reasons why this might be:
- Top sites avoid using video. (Considering these include video specialist like YouTube, BBC and CNN, this is the least likely of the three reasons).
- Top sites avoid using video on the (mobile) homepage. (The homepage of YouTube, for example, is made up of image links to videos, rather than videos themselves. Each video has its own webpage).
- Top sites use video more efficiently (as Dutton suggests).
Querying this apparent anomaly of video usage between all sites and the top 100 with the web performance experts at HTTP Archive, we received the following answer from Rick Viscomi, a leader of the HTTP Archive project and Developer Advocate at Google:
“I think the answer is: efficiency. To be more specific, I think it comes down to autoplay. HTTP Archive just visits a page and records the page load without clicking around. Autoplay videos would be captured on those visits, while click-to-play would not.
“Autoplaying is wasteful for everyone involved because a page visit does not always demonstrate intent to watch. One notable exception is YouTube, where visiting a watch page is definitely intent to watch. Keep in mind that only home pages are crawled by HTTP Archive. So my theory is the top sites choose not to autoplay in order to keep bounce rates low and conversions high.”
Notably, autoplay video and audio is also frowned on from an accessibility perspective. See these BBC guidelines for example. The reason for this is that people with visual impairments rely on screen readers to read aloud a webpage. Clearly if audio or video media starts to play (including advertisements) it will interfere with the screen reader and will make tricky for the user to find out how to make it stop.
The impact of audio on page performance
One of the most useful features of HTTP Archive or WebPageTest (from where it is captured) is the filmstrip which shows how a website loads on a mobile device second by second.
The loading process for New York Times mobile site on May 1, 2017 is captured by HTTP Archive in the image below. The audio story The Daily is at the top of the mobile page, above the fold, allowing us to see clearly how audio may delay page speed.
The audio does not finish loading until 22 seconds, when the play button finally appears and the site is visibly complete.
Poor viewer experience (VX)
Assuming there is no autoplay, a correctly coded website should not require the video to be downloaded until the user requests it by clicking on the play button.
However as soon as the mobile user clicks on that play button, the level of expectation changes…
There are three potential VX problems with video:
- The video is too slow to start.
- It fails to start.
- It stalls during play back – this is due to (re) buffering or a dropping connection, typically shown by the spinning wheel.
- Poor video quality – or quality that is less an optimal for the connection.
Research by Conviva and nScreenMedia (November 2016) illustrates the difference in VX quality when a viewer is indoors (WIFI) or outdoors (cellular) failures for videos to start increases from 1.5% to 2.9% and buffering issues rises from 7.9% to 14.3% of views.
Research carried out by University of Massachusetts and Akamai, of 6.7 million video viewers, in 2012, also shows a growing intolerance to slow, stalling video.
Ramesh Sitaraman, Professor of Computer Science, UMass, Amherst tells ClickZ:
“Mobile users are impatient and abandon videos that do not start up quickly. However, they are more patient than users who have high-speed Internet access (say, Fiber), since their expectations of speed are lower in comparison.
“Mobile users start to abandon a video after waiting for about 2 seconds. By the 10 second mark, if the video has not started, roughly a fifth have abandoned.”
And on stalling:
“We don’t have data split out just for mobile. But, we studied a cross-section of users that included mobile. Overall, people watch videos for a shorter period of time when the video stalls than they would have otherwise.
“Roughly, a 1% increase in stalls leads to 5% decrease in the minutes watched.”
This is Part 2 of a series looking at how video impacts mobile web performance and UX. Read the previous installment: How video impacts mobile web performance and UX, part 1: data and download speed.