For Bjorkman and Zeijlstra (2019), Agree consists of two operations: checking and valuation. Checking involves probing, always upwards from an uninterpretable feature [uF] to an interpretable feature [iF] c-commanding it, and [iF]’s checking [uF]. Valuation generally happens downwards, with the valuer ccommanding the valuee. Upward valuation, in which the probe c-commands the goal, is exceptional and only occurs if downward valuation has failed. In this reply, we argue that this approach is not supported empirically. We present data from Matengo, German, Serbo-Croatian, Sambaa, Liko, and Nez Perce, arguing that upward valuation must be available more generally than Bjorkman and Zeijlstra suggest.