Two main factors that drive up the risk for longterm unemployment are advanced working age combined with decreasing employability. I evaluate a new social experiment which implements a novel active labour market policy intervention in Switzerland that explicitly focusses on the mentioned risk group. It features a fixed treatment plan which combines individually targeted coaching with high-frequency counseling. To increase effectiveness, the interventions happen early in the unemployment spell. The evaluation is based on a unique dataset that merges register data with repeated surveys. Survival analysis shows that the gross effect of the interventions is zero. But this gross result is driven by several overlapping effects, as estimations based on a duration model framework with dynamic treatments reveal. The prospect of being coached slightly diminishes early exits. During coaching, a clearly negative lock-in effect is found. The post-coaching effect depends crucially on whether a search strategy change was recommended by the coach. If this was the case, the lock-in seems to be prolonged; if not, a slightly positive post-coaching effect materialises (these last results are preliminary). In a further step, I will evaluate if salaries in the subsequent job were positively affected.