Amyloid immunotherapy with lecanemab and donanemab has been approved on the basis of statistically significant slowing of cognitive and functional decline in early Alzheimer's disease (1,2). Yet clinicians treating individual patients face the everyday challenge of estimating whether a given patient is declining at the rate that would be expected for them, and whether their trajectory reflects a response to treatment. I argue that this difficulty arises in large part because the field cannot yet routinely stratify patients by tau pathology before treatment, even though tau burden is among the strongest available predictors of both the rate of progression and the magnitude of response to amyloid-targeting therapy. Post-hoc and open-label analyses of the Clarity AD and TRAILBLAZER-ALZ 2 programmes suggest that patients with absent, low, or medium tau burden may constitute a biologically distinct group in whom amyloid clearance is most likely to permit clinical stabilization or measurable functional gain. These observations remain hypotheses, generated largely from subgroup, open-label, and biomarker data rather than from prospective trials designed to test them I propose that tau status should be given strong consideration in patient selection, that tau-guided selection should be evaluated prospectively, and that the access, reimbursement, and standardization barriers to tau positron emission tomography (PET) — together with the promise of scalable plasma tau biomarkers — be addressed deliberately as the field moves toward tau-informed treatment. One tau PET tracer is currently approved by the US Food and Drug Administration, and a regulatory decision on a second is anticipated in 2026.