Abstract The purpose of this study is to demonstrate that the identification of carbonate platforms in frontier exploration using seismic data remains challenging despite the variety of attributes that can be extracted from the data and their integration with other sources of information. A Miocene example from offshore Tobago (Southern Caribbean) is evaluated using 3D seismic data integrated with regional geology, potential fields, analogues and structural restoration. The data are compatible with interpreting the target as shallow-marine carbonates, and a conceptual model is presented. However, far fewer of the seismic characteristics are diagnostic of a carbonate platform, and despite the intensive approach and the use of published criteria for subsurface carbonate interpretation, it was impossible to conclusively identify the target as a shallow-marine carbonate. Alternative explanations such as volcaniclastics, eroded remnants of siliciclastics and basement highs are considered. The study illustrates that it is common to find situations where the origin of a prospective geobody cannot be determined beyond significant levels of uncertainty unless it is drilled. If other elements of the petroleum system are favourable, this irreducible risk has to be accepted to avoid overlooking attractive carbonate reservoirs, provided all available data are used and all possible alternative interpretations considered.