Especially taking into account that knowing some data available via such tests in an earlier stage of life and acting on it could be quite beneficial in a long run.

Whole genome data won't give you health information beyond SNPs because the studies used to generate the health claims use SNPs, because they're cheap.

But there is very little useful data available from SNPs, anyhow. What kind of data do you believe is available and what actions might you take?

If the baseline is a 30% chance of dying of heart disease and it told you that you had a 70% chance, what would you do differently? (but it won't.) Probably you should already be doing it because 30% is a big number!

If the baseline is 0.1% chance of MS and it told you that you have a 0.2% chance of MS, what would you do differently?

The only thing it might tell you that is useful is that you have a high chance of developing what is otherwise a rare disease. It will probably tell you that you have a mildly broken BRCA and have a 22% chance of breast cancer, rather than a 20% chance. That is worthless, even if it is true. But if you're Ashkenazi, there is a 1% chance that it will tell you that you have a fully broken BRCA. That means an 80% chance of breast cancer, which is really not that different from a 20% chance. But it also means aggressive breast cancer at a young age. This is really a different disease. It is a substantial chance of death at a young age. Many women get prophylactic mastectomies. The Dutch have the same 1% base rate as Ashkenazi, but 23andMe does not test for them (at least as of v3). Thus whole genome sequencing can detect the 1% of Dutch with fully broken BRCA and the 1/10,000 of other populations. But those base rates are silly. If you have this gene in your family, you should know already from the high prevalence of breast cancer, especially young fatal breast cancer.

Also, there's a SNP that predicts how fast you metabolize drugs. That sounds useful, but no one takes advantage of it. More generally, it is quite rare that people experiment with doses of drugs, which they really should.

If you do get a whole sequence, you can test every gene to see if it is fully broken ("nonsense mutation" as opposed to the mild "missense"). If you have such a gene, it probably is a big deal, but it will probably be hard to interpret.

The main idea of WGS is having >all< the SNPs whereas you only get the most common SNPs with the SNP tests.

I'm not really sure how you would use the data from WGS (let's say the genome is assembled too - or maybe that would cost more ?). You would probably use BLAT on your local machine to search for genes with known SNPs. I don't think you could do anything more (finding novel SNPs is out of reach here).

I would guess the main idea would be to be able to check for new SNPs as more and more are found in the literature. However, the literature is not that easy to skim through except for the most common SNPs that are already included in the SNP tests.

Going back on the literature: for most multi-factorial diseases, you will see data coming from GWAS and linkage disequilibrium studies that will be really hard to interpret. A SNP popping up like that does not necessarily mean that you've got the trait associated with it.

My comment was probably not really well oriented, but I should still conclude. In my opinion, do a WGS only if you've got enough knowledge of bioinformatics (and I mean an engineer's level). SNP tests are cheap and will provide you with almost everything you could get from a WGS.