Yes, I think that is the experience, for example, in what we called (or call) data science: most of the time is spent in ETLs rather than using ML methods. In a real company linking data difficulty is not technical but time and resource consuming.