The paper addresses the design of adaptive radar detectors with desired behavior, in Gaussian disturbance with unknown statistics. Specifically, based on detection probability specifications for chosen signal-to-noise ratios and steering vector mismatch levels, a methodology for the design of customized constant false alarm rate (CFAR) detectors is devised in a suitable feature plane obtained from two maximal invariant statistics. To overcome the analytical and numerical intractability of the resulting optimization problem, a novel general reduced-complexity algorithm is developed, which is shown to be effective in providing a feasible solution (i.e., fulfilling a constraint on the probability of false alarm) while controlling the behavior under both matched and mismatched conditions, so enabling the design of fully customized adaptive CFAR detectors.