The distinction I see is that an insertion unit only samples a point in the velocity distribution. A full-section magnetic flow tube (Magmeter) actually subjects the entire flow to as uniform a magnetic field as can be generated, and the induced voltage is indicative of the entire integrated flow through the magnetic field.
Magntic flow tubes therefore are forgiving of flow velocity distribution, short piping runs, and other effects that cause assymetrical flow profiles. An insertion meter will only be accurate if you are lucky to get it in the right spot, and it may not be accurate at different flowrates as the velocity profile shifts.
If there are coating problems, Magmeters are available with bullet-shaped electrodes, which may be heated and/or fitted with ultrasonic cleaners.
The magmeter must be kept full as noted in an earlier post. Keep the meter full by either installing it in a vertical line with upward flow, or in a horizontal line P-trap with an offset of anything greater than one [pipe diameter.
One thing that does drive magmeters nuts is if they are close to a chemical injection point and the chemical is still reacting as it passes thru the meter. Many reactions have an electrochemical component, and magmeters work on microvolts, so it is easy to swamp the signal with process noise.
Pulsed DC mags are also more noise immune than the old-style AC mags. The DC meters pulse the magnet coils, when the magnetic field stabilizes they take a voltage reading across the electrodes, then they shut down the coils and check the electrode voltage for a zero value. Then they pulse the magnets again with reversed polarity. Processing these signals lets them filter out biases and quadature errors. The DC pulse rate >can< be faster than line frquency, but it is never coincidental. Variations in local magnetic field (Heavy-current bus-Bar, large AC motor, etc. could cause the same quadrature effects as in the AC magmeter if the DC was pulsing at coincident frequency.