The answer to the question "How can i interpret my multiple sequence alignment?" always depends on the context in which you want to use the MSA.
There are uses of multiple sequences alignments (MSAs) that do not assume that residues in the same column are "homolgous" (a very tricky word, and one I prefer to avoid) i.e. that any differences in the residues in that column are due only to point substitution events. For example, the sequence logos of signal peptides you find on this page
could be thought of as being based on an (ungapped) MSA of a set of not-necessarily-"homologous" signal peptides, but it makes sense to do this as the results/analysis are being interpreted in a structural/functional rather than an evolutionary context.
Having said that, "relatedness" of the sequences in an MSA is an assumption underlying many (probably most...?) applications of alignments. The example above is given just to highlight the importance of understanding the context in which you want to use the MSA.
It all depends on knowing why you're interested in building the MSA in the first place; it can help some people to do this by thinking about which sequences they don't want in their alignment, and then thinking about/understanding why.
An MSA I love to ask students to interpret is the one in figure 2 of this article in PNAS