The parameters modeled range from: (a) Server Side: application’s design, systems architecture, Server SKU’s, geohosting, CDN’s, to (b) Network-Related: bandwidths, connectivity, network latencies, last mile effect, to© Client-Side: OS and browser mix, system configs, to (d) Business-Related: current/projected revenues, market shares, user-minutes, competitive advantage, feature parity, discoverability and adoption, and trends, to (e) Performance-Related: where Hotmail is with respect to its regional/global competitors and previous milestones. The model is fine tuned through AB Testing, comparing the before/after deploying new features, last mile instrumentation, revenue trending and market research.

via Performance-Based Design – Linking Performance to Business Metrics: Velocity 2009 – O’Reilly Conferences, June 22 – 24, 2009, San Jose, CA.

Much like normal web analytics, they look at specific business and market-share metrics to assess their efforts’ success. However, assessing their server and network-related feedback also gives a lot of insight into how their full customer experience is performing and is a good idea for any company using analytics. Remembering to check your user’s actual bandwidth, browser versions, error messages, and so on sounds mundane, but can help you create a positive user experience all around and not just by assessing your conversion funnels.