Yes, I prefer to use two different cutting sites with sticky end.

That also could make sure the direction of the insert is correct.

It may have some advantages, but I remember several years of cloning 7/24 for CD44v K.O. constructs and it appeared to me to be the best to use only blunt end cloning and hundreds of midi- or mini-preps twice a day with of course checking the right direction of the construcs...

EDIT: after about two long years of blunt-end cloning and of course dephosphorylation and usually checking routinely 100-300 minipreps per day to get sometimes just one lucky clone, I was about 6 months ahead of my competitors in the same scientifically highly questionable lab headed by Prof. Herrlich/Prof. Ponta at the former Nuclear Research Center Karlsruhe, Germany - at the end the lab just wasted hundreds of students' years in stupidest competitions.

Different and non-compatible restriction enzymes is best.

If you can use one enzyme with a 5' overhang and another with a 3' overhang, even better. Then you have almost no background colonies.

If you only have 1 site, you can use Gibson cloning to introduce directionality:

https://www.neb.com/products/e5510-gibson-assembly-cloning-kit

I try to avoid blunt cloning under all circumstances. Gibson cloning essentially eliminates the need for blunt cloning.

If you are sub cloning, you might use first, as a cloning vector, something easy to handle like pGEM T, which doesn't need any digestion. Then, to transfer your sequence of interest, use two different restriction enzymes with sticky ends because, like Chun-Hsi Tso said, it will give you also the direction of the insert. I used this strategy several times, succesfully. A thing you may consider when choosing the enzymes is, besides they only cut the MCS and not your sequence of interest, that they are compatible to perform a double digestion, that saves a lot of time and work!

If orientation of the insert is not an issue and find that only one restriction site is your best choice, then, if possible, set the insert:vector ratio to 100:1 or more in your ligation.

Despite using two different sticky end REs you may observe self-religation of the vector. You can do a control ligation where you add water as insert instead of your actual insert. This will show you the ratio of correct ligated to self-religated constructs/colonies. You are then meant to have fewer colonies on the plate that has no insert - but in all honesty this changes for me depending on the backbone vector I am using, e.g. right now I have 22 colonies on my control plate and 4 colonies on my experimental plate. Those 4 colonies all contained a correct insert, whereas the 22 Øcolonies were insert-less. So, in other words, negative ctrl is not always necessary, but do a colony PCR with a cheap polymerase and your sequencing primers to check each colony. This will save you both time and effort avoiding those darn mini preps. PS. You can dephosphorylate the backbone vector and not the insert to increase the effectiveness of ligation. PPS you can transform a very small volume of your ligation to increase transformation (bacteria don't like foreign DNA).