Use them all and allocate spend based on performance (or let CBO do it for you).
They're all different and great to run together.
BUT overall, all roads lead to broad. . .
Which will win over time?
And wait, didn't you just say "it doesn't matter".
The relevant interests and lookalikes can help to target potentially more relevant users early on. If your input is right, those users will convert efficiently.
When those users convert, it teaches FB's machine learning to look for more of those users that purchase.
FB's machine learning learns what types of users convert for each interest and lookalike audience.
It also learns the overall user profile that will convert for anything optimizing to that same pixel event.
That includes *broad* targeting!
Attributes we can't comprehend.
It will also do so more efficiently. . .
This means they can enter more auctions with less competition, which equals lower costs.
So, even if your LAL looks better now, it might not next month as FB continues learning.
This is even more valid now that interests and lookalikes have audience expansion (sometimes automatically turned on).
I already hear someone asking: "If broad is best and everything is basically broad, then why use the other targeting at all?"
Diversity! It can maybe help FB learn more/faster, especially if you have a smaller number of daily/weekly conversions.
Or maybe try going all broad!
Follow me on Twitter if you found this post useful. Let me know if you disagree, or tell me what I got wrong!
You can also subscribe to my newsletter below.