Statistically significant tensions between the Standard Model (SM) predictions and the measured lepton distributions in differential top cross-sections emerged in LHC Run 1 data and became even more pronounced in Run 2 analyses. Due to the level of sophistication of the SM predictions and the performance of the ATLAS and CMS detectors, this is very remarkable. Therefore, one should seriously consider the possibility that these measurements are contaminated by beyond-the-SM contributions. In this article, we use the differential lepton distributions from the latest ATLAS tt¯ analysis to study a new physics benchmark model motivated by existing indications for new Higgses: a new scalar H is produced via gluon fusion and decays to S^′ (95 GeV) and S (152 GeV), which subsequently decay to bb¯ and WW, respectively. In this setup, the total x² is reduced, compared to the SM, resulting in ∆χ² = 34 to ∆χ² = 158, corresponding to a significance of 5.8σ to 13σ, depending on the SM simulation used. Notably, allowing m_S to vary, the combination of the distributions points towards m_S ≈ 150 GeV, which is consistent with the existing γγ and WW signals, rendering a mismodelling of the SM unlikely. Averaging the results of the different SM predictions, σ(pp → H → SS^′) × Br(S → WW) × Br(S^′ → bb) ≈ 9pb is preferred. Assuming that S^′ is SM-like, the 95 GeV γγ excess can be explained if S decays dominantly to W bosons. That latter suggests that S is the neutral component of the SU(2)_L triplet with hypercharge 0.