Well they do say I think that all the Sykeses go back to a common ancestor, give or take. Let's call him Adam Sykes, and let's suppose he had two sons, Bart and Colin.
Bart and Colin inherited Adam's yDNA, approximately, but never quite exactly. If you could compare their full sequences, you'd find differences at a few random points. The trick is to identify those points from the DNA of living testers by dividing them into 2 septs. Doable but not trivial.
The two septs will them mutate independently, and you can repeat the trick to find the first division in the Bartines and the first division in the Colines.
But as you go down the tree, the amount of analysis multiplies, the number of testers needed multiplies, and the number of descendants interested in each finding divides.
Which is why you can't Google up the Sykes family tree as drawn by yDNA. The technology exists, but the economics are impractical.
STRs are alternative DNA. They're God's gift to genealogy - DNA as the testing companies would have designed it. All families have their mutations at the same small set of locations, so you know where to look.
And the beauty of it is, although the SNPs and STRs are quite separate things and mutate independently, the SNP mutation tree and the STR mutation tree must both reflect the same underlying genealogy.
Which doesn't mean they'll be mirror images. For instance, Bart and Colin might both have the same STRs, so the STR tree can't separate the Bartines and Colines.
But then Colin has two sons, Eddie and Freddie, and Freddie has an STR change. So the STR tree can separate Bartines and Freddines. At this point the careless analyst might think he's separating Bartines and Colines - but he isn't, because the Eddines are Colines, genealogically, but are still lumped with the Bartines, STR-wise.
So the STR tree will give a lumped view of the branching, but not lumped the way you'd like. The step-changes in the numbers won't happen at quite the key genealogical points. And the SNP tree, so long as it's incomplete, will also give a lumped view of the branching, but differently lumped. Messy.
(There's massive scope for inferring wrong genealogy here. That careless analyst might well decide that Eddie must have been the son of Bart. It's very easy to turn those STR tables into little trees without realizing that the tree you've drawn isn't the only one you could draw.)