The low impulse is consequence of the low power model and early design being inefficient. Better waveguide designs and increased input will result in increased output. This is the dawn of a technology. Look at the average horsepower of a car in the late victorian times, and now - or the first ford and some common modern model... The first cars would struggle with a hill, or had more than one person - and mechanically fail every 30s - LeMans was impressive, just having a car that could run for 24hrs, let alone racing... over time, through development of the technologies itself, and supporting technologies, this technology became casually realiable... Better usage of the operating principles lead to more tangible output for a given input...
Currently, the tech is looking at least suited for microgravity application... specifically long operation or long distance projects. Not everything requires a lot of thrust, sometimes it's just about putting it in the right place, at the right time... and with EM-Drive, that also opens up for long enough... And you can can parallise - That's per unit. Add more units. If a single satellite can be made to be given viable vector control via such methods the three, maybe six, of these can begin to add together like ants and tow...
Eventually, it's suitable for gravity application, with sufficient development. It's do-able now, really, It's just generating that much energy - that's do-able too - safely is another question, and cheap is definitely another. 80W of microwave energy isn't a lot... most residential models topping 800W.. KW, hundreds of, is feasible. Megawatts if running multiple units for extra thrust is also possible, again not cheap - but it's not impossible to make it work in gravity as is... it would make more sense to make it more efficient before attempting to scale it, tho.