In case that the gold electrode builds a schottky contact, reducing its thickness will degrade such Schottky contact and if nearly removed, there will be no rectifying contact and consequently no pv affect observed.
If you displayed the i-v curve with different gold thickness, and the layer structure of your perovskite cell, one may give more elaborate answer.
Concerning the optimum thickness, it is so that as the thickness increases the metal semicondcutor contact works better since the whole surface will be covered of the metal atoms. In addition the resistance of the metal layer will be smaller reducing the total series resistance of the solar cells and thereby the power loss will decrease and the fill factor increase enhancing the operation of the solar cell. The maximum thickness is limited by the allowed series resistance of this metal layer such that its resistance must be much smaller than the semicondcutor layer resistance.
I think in most perovskite cells the gold does not form a Schottky contact, so I'd disagree with the first comment. It's just an Ohmic electrical contact. So you just want to make it thick enough that you don't have series resistance problems in the gold, and you can calculate that from the geometry of your cell and the resistivity of Au.
The only upper limit on thickess is either practical/price (gold is expensive, depositions take time), or if longer Au depositions degrade your device, e.g. by creating defects and lowering Voc.