An impact and process evaluation showed that although it secured strong local support, improved multi-agency working, increased information sharing and catalysed ‘sensible’ activity; it did not achieve a measurable reduction in violence. This paradox raises a number of important questions: how can multi-agency process improvements be translated into better outputs and outcomes? Can tasking work better in a partnership setting? What can realistically be expected from these schemes? Most critically it highlights the urgent need for more evidence about what works in multi-agency case management and in what circumstances multi-agency case management works.