Drop a single column of measured readings, then enter the LSL and USL from the drawing. The chart bins your data into a histogram and overlays the selected distribution — if the curve doesn't hug the bars, the process likely isn't Normal (try Lognormal or Weibull for skewed data). Cp / Cpk use within-subgroup spread (moving range); Pp / Ppk use the overall spread. A capable process targets Cpk ≥ 1.33. Feed an out-of-spec finding into a 5 Whys or Fishbone to chase the root cause.