Snow thickness on sea ice is a largely under-sampled parameter, yet of importance for the sea-ice mass balance and for satellite based sea-ice thickness estimates and thus our general understanding of global ice-volume change. Traditional direct thickness measurements with meter sticks can provide accurate but only spot information, referred to as "needles" due to their pinpoint focus and information while airborne and satellite remote sensing snow products, referred to as "the haystack" have large uncertainties due to their scale. We demonstrate the remarkable accuracy and applicability of ground penetrating radar (GPR) snow-thickness measurements by comparing them with in-situ, meter-stick data from two field campaigns to Antarctica in late winter/early spring. The efficiency and millimeter- to centimeter accuracy of GPR enables practitioners to acquire extensive, semi-regional data with the potential to upscale "needles" to "the haystack" and to potentially calibrate satellite remote sensing products that we confirm to derive roughly 30 % of the in-situ thickness. We find the radar wave propagation velocity in snow to be rather constant (+/- 6%), encouraging regional snow-thickness surveys. Snow thinner than 10 cm is under the detection limit with the off-the-shelf GPR setup utilized in our study.