Search is an important tool in an ant's navigational toolbox to relocate food sources and find the inconspicuous nest entrance. In habitats where landmark information is sparse, homing ants travel their entire home vector before searching systematically with ever increasing loops. Search strategies have not been previously investigated in ants that inhabit landmark-rich habitats where they typically establish stereotypical routes. Here we examine the search strategy in one such ant, Melophorus bagoti, by confining their foraging in one-dimensional channels to determine if their search pattern changes with experience, location of distant cues and altered distance on the homebound journey. Irrespective of conditions, we found ants exhibit a progressive search that drifted towards the fictive nest and beyond. Segments moving away from the start of the homeward journey were longer than segments heading back towards the start. The right tail distribution of segment lengths was well fitted by a power function, but slopes less than -3 on a log-log plot indicate that the process cannot be characterized as Lévy searches that have optimal slopes near -2. A double exponential function fits the distribution of segment lengths better, supporting another theoretically optimal search pattern, the composite Brownian walk.