ABSTRACT While dockless bike sharing is gaining popularity, oversupplied and poorly maintained bikes introduce chaos and waste (e.g., so-called zombie bikes that unused). Spatiotemporal pattern visualizations can help policy-making and infrastructure improvement (e.g., allocating parking areas). However, multivariate symbolizing (e.g., supply, flow, usage) to optimize dockless bike sharing is challenging. In this paper, we introduce metaphor theory to design multivariate symbols. First, we systemically explore the coupling of three metaphor types (orientational, ontological and structural) with symbols at three levels of iconicity. Then, we construct metaphorical symbols for optimizing dockless bike sharing following a user-centred design process. We also offer an evaluation using eye-tracking and questionnaire techniques. The results indicate that, compared with bin-packing and multiview symbols, metaphorical symbols significantly improved effectiveness and efficiency, and reduced participants' cognitive load. Our evaluation presents preliminary evidence that metaphors can offer new organizational mechanisms for map symbols to represent multivariate naturally and effectively.