As scholarly books remain vital outputs in the humanities, social sciences, and education, new approaches are required to capture their diverse impacts. This study explores the multidimensional evaluation of academic books and textbooks by integrating citation metrics, alternative indicators, and (dis)aggregated library holdings data. Drawing on three large-scale empirical datasets encompassing over 119,000 Scopus-indexed book titles across 26 disciplines, the interactions and divergences among library print holdings (LPH), library electronic holdings (LEH), total library holdings (TLH), and other non-traditional metrics were examined in relation to their capacity to reflect scholarly and educational impacts. Library holdings data are among the most widely available book metrics (covering 97% of titles); however, aggregating print and electronic holdings into a single TLH metric has been shown to obscure important differences. Print holdings, though declining in average count over time, exhibit more stable, cumulative characteristics that statistically align with formal citation patterns, whereas electronic holdings reveal uneven acquisitions and inflated volumes that reduce the predictive strength of TLH. Print holdings outperform electronic holdings in modeling scholarly and educational impacts across most fields, except platforms such as Mendeley and Goodreads, which are better aligned with electronic availability. In a related investigation of textbooks, educational relevance is assessed through Open Syllabus Project rankings and compared with citation-based indicators (Scopus citations and book-to-book citation-based PageRank and HITS rankings), Goodreads user metrics, and WorldCat edition counts. Disciplinary differences emerge as prominent predictors of uptake across metrics. Goodreads ratings provide the strongest predictions in the humanities, WorldCat editions in the social sciences and medicine, and authority scores in citation networks in law and political science, improving the predictability of educational influence. Assessment of scholarly books benefits from acknowledging format-in-use and multidimensional usage patterns. As books often operate across scholarly, educational, and social contexts, their assessments should reflect blended purposes and audiences. Although a book's stated audience and intent may suggest a dominant context, substantial variation exists across titles. This provides a conceptual synthesis to integrate findings from large-scale quantitative studies on library holdings, citations, and altmetrics (alternative metrics) to present a more robust approach for assessing the multidimensional impact of academic books.