Or you can look at the fast-elixir benchmark which breaks down the various techniques by list size.

Enum.concat/2 for two lists is implemented as:

  @spec concat(t, t) :: t
  def concat(left, right) when is_list(left) and is_list(right) do
    left ++ right
  end

Down in erlang, reversing a list executes a heavily optimized native BIF (built-in-function) because it’s a frequently used feature, so it’s probably not as expensive as you think. That being said, unless your lists are long or you’re in a performance critical part of code, don’t worry too much. Do what’s most expressive.

But if you do need to optimize, be sure to benchmark. For some length of list and algorithm append may be faster, but if the length of the list changes, then prepend-then-reverse may be faster.

I find many times the ordering does not have bearing on the correctness of the code. In those cases I prepend out of habit.