Browsers' support
All browsers should be “supported” but the level of support is not universal. Tiered support model is suggested (Fully, Mostly, Partially, Untested) (there is a difference between supporting and testing proactively).
At the top end, the design and functionality should mirror expectations, then with less important browsers we can deviate on design but still offer the functionality, then we compromise the functionality down to base and then there will be a bunch of browser variants that aren’t specifically tests but at the very worst, serving only HTML should still provide base level value.
Regional differences should be taken into account ( for example, QQ browser doesn’t show up much on a global scale, but it can represent 10%+ of traffic in some markets (China/India) etc).
We don’t need to support browsers to a strong extent when their manufacturers don’t support them already (e.g. IE. Base experience only).
Including JAWS into support will require having a license.
Regarding overall browser support, as we are to use progressive enhancement approach with the semantically correct HTML first and CSS on top, then we need to have basic support for most of the browsers.
Regarding screen sizes, we probably should define breakpoints rather than names of the screen sizes. . Screen dimensions are only one attribute.
We need to design with minimal breakpoints and breaking based on the content, not the device.
Please see attached an overview of browser usage by Users & Sessions across the top 20 or so Reckitt sites (representing almost half of all traffic). This is from 1 Sep 2021 to now (almost 6 months). The raw data is there as is a pivot.
Devices' support
Devices' assessment is attached below.
Apple and Samsung dominate, but Xiaomi is also very close behind in third – actually leading in India.