Detailed understanding of the behaviour of an electrocatalyst is not only founded on the bulk properties of a given material, but must also consider contributions from surface structural motifs that often exist on the microscale or smaller. Often, surface features in the form of defects, including atomic step edges or grain boundaries, are considered to be important contributors to the total response of an electrocatalyst when performing a given process, though these are difficult to electrochemically characterise directly due to their physical size. Microscopic electrochemical techniques such as scanning electrochemical cell microscopy (SECCM) offer a convenient means of directly isolating and measuring these types of features, removing ambiguity about their properties and how they contribute to overall electrocatalyst behaviour. In this work, the intrinsic activity of grain boundaries on a polycrystalline platinum surface is investigated for the acidic hydrogen evolution reaction (HER) using SECCM. Through analysing the behaviour of 20 unique grain boundaries, it is established that surface tension effects can have a large conflating effect on apparent activity measured with droplet-based techniques such as SECCM, often producing false positive results in the search for electrochemically active sites. It is only through stringent surface preparation and surface area correction techniques that these features can be meaningfully analysed, with some suggested means of doing so demonstrated. Utilising these procedures, no grain boundaries with activities greatly exceeding that of the surrounding grains were able to be identified, even when working on the sub-micron length scale, suggesting that these sites, if they exist, are rare and would be unlikely to make a significant contribution to the macroscopic HER activity of platinum. Overall, this study presents some common misinterpretations that can result from the SECCM characterisation of grain boundaries and similar features with nanoscale probes, and establishes methods that could be taken in future works to ensure data is truly representative of their intrinsic properties.