The problem that seemed to be largely missed in the discussion yesterday, is that it isn't the lab work that is sloppy (granted, they aren't doing whole genome sequencing, but there is nothing to indicate that their actual lab work is shoddy--edit: particularly when part of their business model is selling their database to pharma). The ambiguity comes in from the analysis and the size of the reference samples and the extrapolation of what we can actually tell from that (a lot less than is commonly assumed). You're not solving the problem by avoiding that part of the equation altogether since that part is the problem. Raw data is largely useless to the vast majority of consumers.

Edit: There is also a reason why the sort of data that you are proposing to provide is usually accompanied by genetic counselling. You might want to consult a bioethicist about your proposal if you decide there is enough interest in the service.