In late 2008 the Weather Channel approached us to design an all new mobile Web experience. Formed as a cable TV channel, they put great stock into analytics, and approached us with specific problems with their existing site, and goals for anyredesign.
They were an easy client to work with in many ways, requiring no convincing at all that everything on the desktop web needed to be on the mobile. Their own stats showed that with their huge user base, even niche products had enough traffic to drive useful traffic, and hence ad revenue.
I was given free reign to develop an all new architecture, driven by user needs based on analytics, and organizational demands for content availability and promotion.
This is a great example of a site far, far too large to design each page. In addition, we were just an agency without a promise of long-term consulting; The Weather Channel would need to maintain it over time.
I developed a modular framework, allowing authoring to mix and match for whatever content was available.
Large audiences mean that not only is all content important, but every type of user and every device has to be considered. In the years before the iPad, I designed for three scales: a featurephone, an old-style smartphone (Symbian, Blackberry) and modern smartphones (iPhone, but also some of those Android things).
These were not hard and fast breakpoints, but used feature detection to load the proper module based on a mix of size, touch, and most of all full javascript support. Back then, many mobile browsers supported script execution robustly, but only at page load.
This attention to scale extended to each module, and in ways much more advanced than even contemporary responsive design allows for.
Data re-flowed, with not just more width and less wrapping, but more actual details available (such as more columns) on larger devices.
In the end, it was a mixed success. We handily met targets and the client was happy with the results. On the other hand, the internal development team was somehow convinced phones didn't support css, so coded mostly as table based layouts. Despite this, our guidance beat the page-size goal (I believe most pages were 14kb or less) and they handily met performance objectives.
The Weather Channel continued to be among the most visited mobile sites in North America, and at the end of this design life, Weather.com was one of the first big organizations to announce they had more mobile traffic than desktop.
It was replaced a year ago by a new design, but parts of it still seem to be used for featurephones, and many of the icons and parts of the structure were adopted for the separately-developed mobile app I otherwise had nothing to do with.
