Figure 10

Modeling sales patterns. (A) Weekly sales of 200 randomly selected bestsellers, with colors indicating best rank on the NYTBL. (B) The weekly sale numbers of the bestselling hardcover The Appeal by John Grisham as reported by NPD Bookscan. (C) The fit on the cumulative sales of (B), empty circles showing the measured sales curve and the green line the model with parameters \(\lambda = 10.37\), \(\mu = 2.03\) and \(\sigma = 1.12\). The fit accurately reflects the data, as evidenced by the high \(R^{2}=0.99\) value. The only exception are the very first weeks where data is often unreliable due to discrepancies between the reported publication date and the book’s physical availability in bookstores. (D) Rescaled sales curve for The Appeal. (E) How weekly sale curves change for different λ, σ and μ values. By tuning these, the many different types of sales histories shown in (A) can be accounted for. Rescaled sales curves of all hardcover bestsellers, (F) fiction and (G) nonfiction. Most book sales curves closely follow a single formula, indicating that the model captures the patterns correctly as given in Eq. (5)