Comparative studies of generative space-planning methods are most useful when they move beyond an overall ranking and explain how performance changes across recurrent spatial conditions. This article presents a secondary, typology-conditioned reanalysis of a 50-case benchmark for adaptive hotel planning in existing buildings. The benchmark evaluates three algorithmic strategies for room allocation within fixed floor boundaries: a general-rules tessellation method, a self-organization method, and a mixed corridor-attraction method. The underlying benchmark used a common brief with a target room area of 25 m2 and corridor width of 1.6 m, and it assessed outcomes across I-, L-, C-, T-, and H-type plan families. The present study reanalyzes the published case-level reviewed room counts for all 50 benchmark cases and treats reviewed capacity as the primary outcome because it reflects post-validation usable yield under a common proof-of-concept brief. The article augments whole-sample comparisons with typology-specific descriptive statistics, repeated-measures nonparametric tests, pairwise dominance analysis, winner-share analysis, a leave-one-case-out sensitivity check, and a detailed appendix of case-level rankings. Across the full benchmark, the general-rules method produced the highest reviewed total (1588 rooms), followed by the mixed method (1468) and self-organization (1307). The overall difference was statistically significant (Friedman χ2(2) = 50.51, p < 0.001, Kendall’s W = 0.505), the same hierarchy persisted in each typological subset, and the pooled ordering was preserved in all 50 leave-one-case-out re-estimations. However, the margin of superiority varied substantially. The mixed method remained locally competitive in several I-, C-, L-, and T-type cases, whereas the general-rules method was especially dominant in H-type plans. These findings show that typology-conditioned benchmarking yields a more operational basis for algorithm selection than pooled comparison alone, while also indicating that the evidence should be interpreted within the reviewed-capacity criterion used by the source benchmark. For early-stage adaptive hotel planning under that criterion, tessellation should be treated as the default baseline, the mixed method as the principal flexible alternative, and self-organization as an exploratory supplement rather than a primary capacity-maximizing routine.