The long-standing issue of hydrological predictions in ungauged basins has received increased attention due to the recent IAHS Decade on Predictions in Ungauged Basins (PUB) initiative. Since the outset of PUB, many have noted that the best way to confront an ungauged basin is to first make some basic streamflow measurements. In this study we explored the value of a rudimentary gauging campaign for making predictions in an ungauged basin. We used the well-studied Maimai watershed in New Zealand as a hypothetical ungauged basin where we pretended to start with no runoff data and added iteratively different sub-sets of the available data to constrain the calibration of a simple three-reservoir conceptual catchment model. These subsets included single runoff events or a limited number of point values - in other words, what could be measured with limited, campaign-like field efforts in an ungauged basin. In addition, we explored different types of soft data to constrain the model calibration. Model simulations were validated using the available runoff data from different years. We found that surprisingly little runoff data was necessary to derive model parameterizations that provided good results for the validation periods, especially when these runoff data were combined with soft data. The relative value of soft data increased with decreasing amount of streamflow data. Our findings based on the Maimai watershed suggest that when starting with no flow information, one event or 10 observations during high flow provide almost as much information as three months of continuously measured streamflow for constraining the calibration of a simple catchment model.