This paper reports an empirical audit of design consistency across 47 production no-code software products (2023–2025), comparing products built with a documented, tool-enforced design-token system (n=19) against products styled ad hoc, on a per-element basis (n=28). Two independent evaluators scored 10+ screens per product on a 0–100% conformance scale across four visual dimensions—color, typography, spacing, and border-radius—using a standardized heuristic rubric (inter-rater agreement κ=0.87). Token-based products scored 94% overall consistency on average versus 38% for ad-hoc products, a 2.5× difference, with the largest gap in spacing conformance (97% vs. 38%). We report the rubric, the score breakdown by dimension, and discuss why spacing appears most sensitive to the absence of a systematic base unit. We situate the finding against the W3C Design Tokens Format Module and discuss the audit's methodological limitations, including sample-selection transparency, evaluator independence, and the correlational (non-causal) nature of the comparison.