This paper studies a seller's compound search for bidders by a deadline. We show that the optimal search outcomes can be implemented by a sequence of second-price auctions, characterized by declining reserve prices and increasing search intensities (sample sizes) over time. The monotonicity results are robust in both cases of short-lived and long-lived bidders. Furthermore, a seller with short-lived bidders sets lower reserve prices and searches more intensively than one with long-lived bidders. We also show that the inefficiency of an optimal search auction can stem from its inefficient search rule.