By Aaron Rudger | December 20, 2012
Earlier this week, I had the privilege of attending the San Francisco/Silicon Valley Web Performance Meetup for a talk on mobile web performance by Estelle Weyl. The talk was fabulous, with deep discussion on web performance from 4 angles (limitations) that are unique to the mobile experience:
- Power– Mobile devices lack a continuous source of power
- Latency– Mobile devices depend on extremely latent wireless connectivity
- Memory & processing– The memory & processing power afforded to browsers in mobile devices is tiny
- UI responsiveness– Touch
Each one of these factors weighs heavily on a user’s mobile website experience and is most commonly unappreciated and/or misunderstood by developers and site owners coming from the desktop environment. These differences can be on the magnitude of 2-100 times(lower for mobile).
And yet the habits (leaving multiple applications open and running, navigating freely from site to site) and expectations (fast pages, beautiful design and rich interactivity) of users aren’t really any lower than their desktop experiences.
It’s amazing that people use their smartphones to browse the web at all!
Credit the amazing capabilities of the Webkit browser, the growing availability of WiFi and some pockets of really good innovation. But the reality is that the mobile web still has so much untapped potential!
So where do you start to ensure you deliver a delightful mobile web experience to your customers?
A good place to begin is with metrics. The meetup discussion was intentionally absent of metrics, and that’s OK. Gathering performance metrics can be difficult. Often, testing basic functionality and raw usability across multiple devices is a much bigger fish to fry. But Keynote offers a free tool that can be helpful: MITE. It tests the performance of content and visually shows you rendering, across devices using emulation. Another great option is Webpagetest which incorporates latency into its timings. And soon, we’ll be offering a free real device testing service for functional testing as well—you can request an invite for DeviceAnywhere FREE, today.
But to really understand the nature of wireless carrier latency, there’s no substitute for continuous monitoring, from the end user perspective. Here’s a pretty dramatic example, taken from 1 week’s measurements of a top sporting goods retailer’s web and mobile sites:
Notice the distribution of the mobile measurements (wide curves are bad). And that’s just from one device, over one carrier. Think about the diversity of your real users’ devices and connections! The “long tail” effect common with web measurements tends to be greatly exacerbated by the latency experienced by mobile devices. This may not be revealed by simple testing. The breadth of degradation makes it imperative to understand performance over the air.
Estelle wore a t-shirt to the meetup emblazoned with “Got Latency?” Your growing base of mobile customers surely do! Keep this at the top of mind as you continue your journey with the mobile web.