What is the accepted level of text analytics accuracy?

The most basic open-source sentiment analysis models, trained on very general sentiment libraries, achieve around 70% accuracy.

Commercial industry-specific analysis tools, such as BytesView or MeaningCloud, can achieve up to 85 percent accuracy or higher.

As sentiment accuracy is determined by a number of factors, such as the sort of data you're operating with, the people who are tagging your sentiment database, the complexity of the text used in your industry, such as finance, healthcare, or pharmaceuticals data, and so on.