One of the first things the company should probably do is to reorganize the first page that potential customers see. They need to make sure that potential customers have a good idea about what they can and can't get from these sorts of genetic tests.
For example, right now a prominent feature is a box that lets the viewer search for the diseases 23andMe “covers” along with a list of popular topics. People may come away thinking 23andMe has useful tests for most of the diseases listed. They don't.
They have some useful tests for a few, rare genetic diseases. But the bulk of their tests are not at all useful yet in figuring out someone’s risk of getting a certain disease. What they have for the more complicated diseases is a way for people to compare their DNA to various studies in the scientific literature.
Maybe a study was done that found a DNA difference involved in diabetes. Customers can see whether or not they have this difference too but this tells them nothing about their risk for diabetes. It gives just one piece of a giant puzzle. They are not getting any meaningful results that can predict their risk for diabetes. This box should probably be heavily modified or even eliminated.
In fact, the website really should be organized into different sections that are labeled by how medically useful they are to the customer rather than by how strong the DNA study was scientifically. Maybe they could split their tests into three sections.
The first would be carrier testing. These tests can tell you if you have a hidden genetic disease that you could pass down to your child if your partner has it as well. This would get high marks for reliability, scientific validity, and usefulness.
The next section would be more fun related stuff. This would have ancestry and some of the traits testing. It would be able to tell you what your earwax is like, where your mother's, mother's, mother's, etc. mother came from, the odds that your child might have blue eyes, etc.
The final section would include the bulk of what is tested. These are the tests that compare your results to results in the scientific literature for complex diseases. Many of these tests would score high in scientific validity but get no points for usefulness. As I said before, most if not all of these tests will not give you an accurate risk assessment for the diseases they look at. Period.
There isn’t any reason these results shouldn’t be included, though. Maybe people enjoy seeing the results or want to use them to watch progress in the field or whatever. But the companies need to say upfront that these tests are not that useful for determining risk. This needs to be obvious enough that someone wouldn't buy the product just for that test.
As a last point, 23andMe (and all genetic testing companies) need to be much more upfront about what their tests can offer based on race. The carrier tests are probably pretty good for most everyone (although they may miss any nonwhite versions of many diseases). The fun section might be pretty useful to the nonwhite world for ancestry but probably less so for traits as they have mostly been determined for people of European descent.
Most of the rest of the tests they offer that deal with more complex diseases have only been validated for white people. This needs to be explicit on their website so nonwhite people know they aren't getting as much bang for their buck. Buyer of color beware!
These kinds of changes will go a long way towards making these sites more transparent to potential customers. And they may even keep the FDA at bay.