Ok, I ran some calculations and created two usefully helpful tables of cost per damage efficiency of AP/AD and penetrations.
How to use them:
On the X-axis you find the defensive stat of your opponent. Here you have to consider the effective armor/MR AFTER the OTHER reductions you already have, in order to decide whether you should buy AP/AD or penetration.
On the Y-axis you find the scaling to current damage ratio. Remember to include all of the current damage to calculate the scaling-to-damage (StD) ratio. Exemple: I have 100+80%AP and I already have 15 AP. the ratio is 0.8/(100+0.8*15) =0.00714.
Find your right square, check the scale and if the scale says 1 or more buy AP/AD, if less than 1 buy penetration.
Ex: still with an StD of 0.00714, if your target has 50 MR you look at the table and see tht you re in the light-blue area. On the scale that means a value between 1.5 and 2, meaning you should buy AP and not penetration to increase your damage efficiently.
NB: I have considered Void Staff as the penetration item above 65 MR and LW as the penetration item above 200 Armor, while After 200 effective MR I consider flat penetration again, as you should have bought a void staff long ago... These are roughly the points where scaling penetration becomes more cost+damage efficient than flat penetration.
Efficiencies of AD and AP vs penetration by Shinaliss
I have to say that the gold-efficiency of the Brutalizer is surprising and looking at these plots I would advice it to any AD champion, even carries, maybe getting a ghostblade afterwards.
I am here to answer any question. I want to post this on the Cost-Efficiency page, but I wait for your comments before doing that.
EDIT: since some of you asked for a simpler plot showing a line above which one should buy AD/AP, I produced that too. Of course it gives you no clue on how much you are wasting by buying the "wrong" stat, that in some cases could be useful to obtain some big and highly cost-efficient items.
Border of AP and AD efficiences vs penetration items by Shinaliss
Cheers,
Shinaliss