I am fairly sure this is a duplicate, but surprisingly I could not find it. Feel free to redirect me and close the question...
I am currently trying myself in the dark art of computing spectral sequences and am quite confused about the indexing conventions. If $\Phi_\ast$ are the graded abelian groups we care about and $X_s$ is our (co)filtered object, many introductions start motivating spectral sequences with the grading $\Phi_n(X_s)$, but as soon as they introduce things like exact couples they switch to Serre-grading $n=p+q, s=q$.
Why?
Is it just for historical reasons? Is it just to avoid some signs? Are there empirical reasons like "it often happens that this way the data becomes more aligned with coordinate axes" (and if so, are there good reasons for this empirical evidence)? Or does the Serre grading naturally appear in some construction I am not aware of yet?
Also: if the Serre grading makes more sense than the naive grading, why are there instances where the Adams grading is preferable? And are there even more grading conventions I am not aware of yet (the difference between homological and cohomological grading does not count)?