I'm asking about the conductor of an elliptic curve. I'm looking at the definition given as $$N_{E/\mathbb Q} = \prod_{p} p^{f_p},$$ and I understand that $f_p$ encodes information about the ramification of $p$ as follows: $$f_p = \begin{cases} 0 \text{, if $E$ has good reduction at $p$,}\\ 1 \text{, if $E$ has multiplicative reduction at $p$,}\\ 2 \text{, if $E$ has additive reduction at $p$, and $p\neq 2,3$,}\\ 2+\delta_p \text{, if $E$ has additive reduction at $p=2\ or\ 3$.} \end{cases} $$
Above definition from planet math.
I think one could say the conductor is a product which represents which primes the curve has bad reduction, and how bad it is.
What I don't understand is how we go from this construction to considering it so important that databases like Cremona's and the LMFDB are principally ordered by conductor.
I have noticed that computing with low conductor curves is faster. Is this true? Is there a reason why low conductor would imply computational convenience when doing elliptic curve arithmetic? Does low conductor imply that the $a$-invariants will be small, or even bounded in some way?
Elliptic curves of large rank and small conductor by Noam Elkies and Mark Watkins is a cool paper I've been skimming, but I wish I knew why finding curves of low conductor was so motivating.
Thanks.