I advise caution about various percents being thrown around without enough context. I understood the 50% figure above as describing the conversion of heat energy into motion, e.g. inside of a natural gas power plant. The figures you quote are defined as being on the customer side of the energy meter.

Further below that calculation is updated to try and reflect the source fuels using some US EPA numbers: "The (US averaged) inefficiencies recalculated relative to source fuels energies are hence 25% for induction cooking surfaces using grid electricity, 84% for induction cooking surfaces using on-Site Solar, and 38% for gas burners.

The original point (that burning fuel to heat things to spin things to make electricity which is moved to your home to heat a pan is less efficient than moving the fuel to your home and burning it there to heat a pan) is maybe better illustrated by the EPA source-site ratios given to make that adjustment: "3.34 for electricity purchased from the grid, 1.0 for on-site solar, and 1.047 for natural gas. The natural gas figure is slightly greater than 1 and mainly accounts for distribution losses". So if the "electricity purchased from the grid" was generated with natural gas in the first place, you can and see the difference there.