I almost always see explanations of monomorphisms such as: if $\forall g_1,g_2, \ m\circ g_1 = m\circ g_2 \Rightarrow g_1=g_2$, then $m$ is monic (and similarly for epimorphisms). I find this is not at all intuitive.
Is it OK to turn the argument around and talk about inequality between arrows instead of equality? (btw, this is for a blog post, not for a scientific paper)
Monomorphisms:
$m$ is monomorphism if does not erase information: if $g_1$ and $g_2$ are different arrows, we can still tell them apart even if they are followed by $m$: $$ g_1 \neq g_2 \quad \Rightarrow \quad m \circ g_1 \neq m \circ g_2\quad (\forall g_1,g_2) $$ In other words, the fact that $m \circ g_1 = m \circ g_2$ can't be because $m$ "hides" the differences between $g_1$ and $g_2$, it really must be because $g_1 = g_2$.
Epimorphisms:
$e$ is an epimorphism if it does not restrict access to information: if $g_1$ and $g_2$ are different, we can still tell them apart if they are preceded by $e$ (e.g. $e$ does not hide the parts of the source where $g_1$ and $g_2$ would have differed): $$ g_1 \neq g_2 \quad \Rightarrow \quad g_1 \circ e \neq g_2 \circ e\quad (\forall g_1,g_2) $$ In other words, the fact that $g_1\circ e = g_2 \circ e$ can't be because $e$ "sets things up" so that $g_1$ and $g_2$ look equal, it really must be because $g_1 = g_2$.
This is fine. Your definitions (which are the contrapositive of the conventional definitions) are logically equivalent to the conventional definitions, and if you think your definitions better supports an intuitive understanding then you are free to use them.
I would add a footnote explaining that this is not the conventional way of writing down the definition, just so your readers are prepared should they want to read up on it on their own afterwards.