Description
Collect load times from real visitors, segment them by location, network and device, report median and 95th percentile, and track them over time to find and explain regressions.
Synthetic tests measure a page in controlled conditions; real user monitoring captures timings from actual page views, so it shows the speed people really experience. The team identifies its major pages (home, top landing pages, checkout and other high-traffic areas), segments their load times by geography, network type and device, and reports both the median and the 95th percentile, which captures the slowest 5 percent of views and mostly reflects slow connections and devices. Gaps between segments and pages are explained and used to prioritise fixes. Results are tracked week over week and quarter over quarter, with the reason for each change logged in a report or wiki, dashboards marking performance budgets, and alerts to page owners when timings move. Because averages hide the tail, percentiles are essential; Google Analytics-style averages are not enough.
Attributes
- Discipline
- Product Analytics & Strategy
- Loss type
- Siloed
- Stage
- Post-launch
- Data
- Quantitative
- Effort
- Days
Steps
- Choose a RUM tool that records timings from real page views and reports percentiles.
- Identify the major pages: home, top landing pages, checkout and other high-traffic areas.
- Segment load time by geographic location, network type and device.
- Report median and 95th percentile load time for each page and segment.
- Compare segments and pages, isolate causes of gaps (including experiment variants) and prioritise fixes.
- Track week over week and quarter over quarter, logging why each change happened, with dashboards showing budget lines and alerts to page owners.
Sources
- Designing for Performance (Hogan, 2014)
Also known as
- RUM
- Field performance monitoring
- Real user monitoring
Categories
- Analytics & Behaviour
- Product Analytics & Strategy