The first, and most important, thing to remember is that ABO antigens are "carbohydrate-based" and are not, therefore, direct gene products (not that any antigens are, as every one of them undergo post-translational changes). The direct gene products are, of course, the A, B and H transferase enzymes. At birth, it is incredibly rare for the enzymes to be "working" at its optimum/maximum, so that it is rare for the ABO antigens to be expressed maximally (or anything like) at birth. I am certain that you know all this already, so that I am probably "teaching my Grandmother to suck eggs", as the old (and in this case, almost certainly, insulting) adage goes.
As a result of the above, however, unless you can perform A, B and H typing by molecular techniques (NOT to be recommended - see Geoff Daniels book, Human Blood Groups), you either have to decide to ignore all serological cord ABO types, and call all of them O, or, you have to use serological methods that will enhance the antibody/antigen reactions. Herein, there are inherent problems.
Firstly, whatever enhancement you use, you MUST use a suitable negative control. It is fine (in my opinion) to vary the incubation temperature from RT to 4oC, but, to so do, it is very necessary to use another cord blood from a known group O cord sample (i.e. where both parents are KNOWN to be group O themselves, and so an A or B subtype in terms of the control is not a problem).
Similarly, the same can be said for enzyme-treating the baby's red cells, as long as the control cells are also treated in EXACTLY the same way with the proteolytic enzymes.
Finally (at least for now!!!!!!!), it should be remembered that we routinely use monoclonal ABO antibodies these days. These are extremely avid, which is fantastic, but are also VERY specific, which can be a drawback. By this I mean that the old polyclonal human-derived ABO antibodies we used to use (when I was middle-aged, and Karl Landsteiner was a young boy) had the single (and probably only) advantage that they were not quite so specific, and would, therefore, detect ALL (or most) ABO antigens, including those that the monoclonal antibodies would not necessarily detect. For an explanation of this, there was a recent paper in Vox Sanguinis (Cripps K, Mullanfiroze K, Hill A, Moss R, Kricke S. Prevalence of adsorbed A antigen onto donor-derived group O red cells in children following stem cell transplantation: A single-centre evaluation. Vox Sang 2023; 118: 153-159. DOI: 10.1111/vox.13386) talking about the A antigen being adsorbed onto the surface of group O red cells in vivo. One of the references they use is the first peer reviewed paper that I ever wrote, concerning A and/or B substance being adsorbed onto the surface of donor-derived red cells in vivo. What I failed to say in this paper was that this phenomenon was far easier to detect with polyclonal ABO reagents than monoclonal ABO reagents (36 years, and I still regret this omission!).
Anyway, IF I HAVEN'T SENT YOU TO SLEEP YET, my point is that, as long as you use suitable controls, particularly NEGATIVE controls, there is no reason why you should not use any modification to any technique (GIVEN THAT IT IS IN YOUR SOP, with all the qualifications given above), and, even then, if you feel it safer, GIVE GROUP O BLOOD.