Experimental data have provided intriguing hints for the violation of lepton flavor universality (LFU), including B → D(^∗)τν/B → D^(∗)ℓν, the anomalous magnetic moment of the muon and b → sℓ⁺ℓ^- with a significance of >3σ, >4σ, and >5σ, respectively. Furthermore, in a recent reanalysis of 2018 Belle data, it was found that the forward-backward asymmetry (ΔA_FB) of B → D^∗μν¯ vs B → D^∗eν¯ disagrees with the SM prediction by ≈4σ, providing an additional sign of LFU violation. We show that a tensor operator is necessary to significantly improve the agreement with data in ΔA_FB while respecting the bounds from other b → cℓν observables. Importantly, this tensor operator can only be induced (at tree-level within renormalizable models) by a scalar leptoquark. Furthermore, among the two possible representations, the SU(2)_L-singlet S₁ and the doublet S₂, which can interestingly both also account for the anomalous magnetic moment of the muon, only S₁ can provide a good fit. Even though the constraints from (differences of) other angular observables prefer a smaller value of ΔA_FB than the current central one, this scenario is significantly preferred (nearly 4σ) over the Standard Model hypothesis, and is compatible with constraints such as B → K^∗νν and electroweak precision bounds. Therefore, if the ΔA_FB anomaly is confirmed, it would provide circumstantial evidence for scalar leptoquarks and pave the way for a natural connection with all other anomalies pointing toward LFU violation.