This study examines "market size inversion": where novel innovations succeed despite initially low market size projections, and traditional innovations fail despite high projections. Contrasting the strategy literature's supply-side competition-centric view, this paper introduces a demand-side model. It posits that because customer evaluations of novel innovations are heavily influenced by early adopters, a significant portion of demand is observable only post-diffusion. Therefore, pre-launch observable demand for novel innovations is downward biased. Utilizing agent-based diffusion simulations and empirical analysis of over 33,000 consumer product launches, the results confirm that novel products outperform non-novel ones with equivalent pre-launch observable market sizes. Further analysis demonstrates that successful innovation selection by firms involves balancing novelty with observable market demand. This paper contributes to the strategic innovation literature by offering a demand-side perspective on the value of novel innovations, and articulating the limits of data-driven innovation.