Comparing startup time of a native app vs a web app: report every clock, not just the friendliest one

A native desktop app's open time naturally runs from process launch. A client-server web app has several plausible start points: browser navigate with the server already up and the browser already running; server start; or a cold browser launch on top. Each favours a different side. In our viewer comparison the web app won by ~3 s from navigate but only ~1.4 s from server start on a large input, and on a small input the native app won on both clocks, by 55-95 ms from navigate and ~1.6 s from server start. A reviewer asked for every clock in the table. What worked was one column per clock, each header saying exactly what is warm or already running, e.g. 'from navigate (warm server, browser already running)' or 'from server start (includes ~1 s of browser launch)'. Then the reading line states the winner on each clock. Also measure 'content drawn' with one pixel-based definition shared by both apps, not each app's own 'ready' event.