Abstract: A method of evaluating an inline inspection recipe compares the capture rate of metal pattern defects in bounding boxes arising from failed electrical test vectors to the capture rate after the bounding box is shifted. A difference between the first and second capture rates indicates whether the inline inspection recipe is valid for capturing killer defects, or if the inline inspection recipe needs to be adjusted. In a particular example, the electrical test vectors are directed at a selected patterned metal layer of an FPGA (M6), and the metal pattern defect data for the selected patterned metal layer is mapped to the bounding box determined by the electrical test vector.