When YDNA test results are posted they usually show data from multiple test subjects sharing a common surname. The results are given in terms of values for each marker tested for each test subject. Test results are typically posted for a certain number of markers: 12, 26, 36, and 67 markers are commonly tested, but in some cases more may be tested.
The underlying idea for posting these test results is so that people can find genetic "Matches" between persons sharing the same surname. If there are MANY test results, (many different test subjects), showing results for a large number of markers, the amount of data that needs to be processed to find a "match" can be very large. Finding a match under these circumstances can be an onerous chore.
One way to simplify this is to develop a "Modal YDNA signature" for the test results, or some subset of those results. The signature is defined as the "mode" (typical value) for each marker across all test results in the subset of the data being considered.
Here's a fairly simple example, based on the Snoddy YDNA test results. The Snoddy YDNA website shows that they have tested for only 12 markers. Because the name is relatively uncommon, there are only six reported test results.
Note that in this example most of the test subjects (four of six) match the modal signature exactly. Because of limited number of markers tested, this may or may not say much about how these four persons relate to each other. They may share a relatively recent common ancestor, or perhaps not, as these results could arise simply by chance. (To evaluate this you'd need to test more markers. Typically, a 36 marker test gives good, but economical results. Increasing the number of markers tested to 67 might yield somewhat better results, but that improvement comes with increasing cost, and may not be cost effective.)
The other two test subjects give us somewhat more interesting information to talk about. One differs from the modal signature by one marker, and the other by two markers. If you compare these two test results to each other (not to the modal signature) they differ by three markers. A mismatch of three out of twelve markers (3/12 mismatch, or "three off in a 12 marker test") is good evidence that these two individuals do NOT share a common ancestor in anything like a meaningful timeframe. This despite the fact that the participants identify the same individual (Person:John Snoddy (4) in relate's system), as their ancestor. To be blunt about this, at least one of these test subjects is wrong about their ancestry. With 25% of their tested markers differing, they do not share a common ancestor in a meaningful timeframe. One of them might have it right, but we can't say which one, based on these data. But at a minimum, these results suggest that the two test subjects need to re-evaluate their lineage.