My first results are not that bad considering the messy translation memories I used.
I decided to make life difficult for Slate Desktop and to feed it a real-life scenario of messy TMs. Still, my engine created exact matches of my own work almost 30% of the time, and that was seven times greater in numbers than Google Translate. Slate Desktop calls this the Quality Quotient and claims mine is within the average user’s experience. In general, my first results are not that bad, considering the messy TMs I used. Slate™\Desktop allows me also to force terminology, if I want. Review: Who Is a Translator’s New Best Friend? and Practical MT Evaluation for Translators