As the adoption of residential photovoltaic (PV) continues to increase, its influence on distribution feeder voltage and currents also increases. Effective allocation or modelling on the appearance of PV across the system as a function of an adoption forecast is an important consideration for future distribution planning. The spatial information required for the forecast/allocation process is expected to be available to utilities at a cost proportional to its level of detail. Naturally, there is a need to understand the potential trade-offs between different modelling approaches and options. This paper explores three allocation models that differ on the complexity of their allocation mechanism. Both the error due to ignoring customers' PV adoption mechanisms and due to the PV forecast uncertainty are explored and compared.