Imperfect detection of individuals in threatened wild populations is common and can obscure real population trends when it is unaccounted for in population monitoring, and therefore impede conservation decision making. For many threatened insects, there is a lack of biological information or available long-term data to inform how best to practice data collection and population monitoring. Here, we inform the design of a long-term population density monitoring protocol for Brachaspis robustus, a Nationally Endangered grasshopper endemic to the Mackenzie Basin of New Zealand. We use removal sampling (repeated visual searches of a predefined area where any individuals found are temporarily removed to achieve successive depletion) during a single austral summer season (November to March) to rapidly quantify seasonal and demographic visual detectability. Juvenile instars dominated population composition in all months except December and males represented > 50 % of monthly captures. Adult females were 2-3 times larger than adult males, and 79 % of those captured were found during the first search of an area compared to only 52 % of adult males. The odds of detecting an individual increased by 6 % per 1 mm of body length. Removal sampling was found to be an effective method for rapidly informing future long-term monitoring design for a visually cryptic, threatened insect. Recommendations include monitoring adult females as an index of population size, restricting monitoring to when adult abundance peaks (November and December), and conducting multiple monitoring events within peak months to counter the effects of daily and seasonal variation and imperfect detection.