Most of the risk in a coastal project is decided before any design work starts, in the quality and coverage of the survey behind it. Bathymetry, topography and ground conditions set the boundaries within which every later decision is made, and an assumption carried forward from a thin dataset tends to surface later as a variation.
Bathymetric coverage is the usual weak point. Single-beam echo sounding along widely spaced lines leaves the interpolation to do work it cannot reliably do, particularly across a reef edge or a dredged channel where the bed changes sharply between lines. Multibeam coverage costs more in the field and removes an argument later.
Ground investigation is the second. Reclamation settlement, quay wall stability and pile capacity all depend on what is beneath the seabed rather than on it, and boreholes placed to suit access rather than the structure produce a design based on the wrong locations.
The point of survey is not to produce a drawing. It is to reduce the number of things still being assumed by the time the design is committed, and to make the remaining assumptions explicit.