When an action predictably yields a contingent effect, we associate action and effect, allowing us to proactively shift our eyes to the effect’s anticipated future location. Such anticipatory saccades prepare for the later comparison of expected and actual effect (proactive effect monitoring). Importantly, anticipatory saccade latencies are adapted to the future effect’s timing. When an effect predictably appears after a short(long) delay, we initiate the first anticipatory saccade towards it earlier(later). Here, we used this temporal adaptation of anticipatory saccades to the timing of future effects to assess how relevant as compared to irrelevant effect features are proactively monitored. Effects of left(right) responses predictably appeared after a short(long) delay. In one half of the experiment, only action-irrelevant effect features (coloured circles) were visible. In the other half of the experiment, a relevant effect feature (number to respond to) appeared within the irrelevant effect some time after its onset. Crucially, the short(long) delay until the onset of the relevant effect feature was manipulated inversely to the long(short) delay of the irrelevant effect feature. We observed that participants’ first anticipatory saccades occurred earlier(later) based on the delay of the irrelevant effect feature which always occurred first. The timing of the subsequently-occurring relevant effect feature did not yield an influence on anticipatory saccade latencies. Thus, we conclude that the timing of proactive effect monitoring is determined by the temporally first effect features rather than the most relevant effect features.