NLP is arguably very difficult to analyze, because it's not a single body of coherent knowledge forming a model, rather than a mash-up of psychological techniques and some assertions about how the minds works.
I think that when you can extrapolate something that is definitely an assertion about the mind or how some techniques improves the life of people who use it, then you can test it. And it's usually found to be false.
There are however some assertions that turned out to be 'true' (that is, an experiment showed some effects), like the mirroring effect, and others that were borrowed from other fields or experiments.
It's better not to be too hang-up about the pseudoscience label: just know that when you talk about NLP, you are entering in a field of not necessarily related beliefs which are mostly false.