23andme shows a man as my dad's cousin, but ancestry.com shows him as my dad's brother

Question

I did 23andme, and so did our mystery man. Two of my grandfather's brothers also did 23andme, as well as a number of the kids and grandkids of my grandfather's brothers and sisters (my grandfather had 9 siblings).

23and me shows this man as my dad's cousin (my first cousin once removed), and the man's number (8.00% DNA shared) is entirely outside the range an uncle would be per their website (17%-34%), and entirely inside the range for cousins once removed (2% to 11.5%). This man's number is consistent with many of my other known first cousins once removed on 23andme (all in the 6 to 8% range). So far, so good.

But this man and my grandfather (now deceased) both did Ancestry.com (my grandfather did it many, many years ago). That website shows this man to be 100% certain to be the son of my grandfather, so my uncle (my dad's brother), with 3480 cM across 25 segments.

There is the question. One website seems to rule my grandfather out as this man's father, the other seems to rule him in. There is a conflict, and I don't understand how this can be.

Answer 1

Are you sure your mystery man is the same person on both sites? How do you know?

8% shared DNA (as per the 23andme results) could be related to your Dad as:

(figures from https://dnapainter.com/tools/sharedcmv4/597)

As you say, 3480cM is almost certainly the son of your grandfather.

So your first step should be to attempt to verify that the 'mystery man' on Ancestry and 23andme is the same person.

Answer 2

First, as already indicated, you need to be sure they're the same person.

Second, you need to compare apples & apples - although 23AndMe made the unfortunate choice to show matches primarily in %, they also show cM (total and segments), which is the standard. On Ancestry, you may need to click through the match numbers to see the pre-Timber match levels. If the cM numbers don't match in both places (or very nearly), then something's wrong.

Third, you can't rely on either site's suggestion as to the nature of the relationships, as they may be overly general or optimistic. You need to use a common reference source - the Shared cM tool on DNAPainter.com is generally recommended. Many match levels correspond to more than one potential relationship, so you may also need to look at the ranges for each possibility.