Dispatch rebuilds the schedule every morning. That is not the optimiser failing. It is being fed job durations and travel times nobody has measured.

Almost always because the optimiser was fed inputs it could not honour. It schedules exactly what you told it, so if travel time assumes empty roads and service durations came from a product sheet rather than a stopwatch, it produces a day nobody can physically complete. Dispatch overrides it once, then twice, and within a month the schedule is manual again and everyone concludes the optimiser does not work. We measure durations and travel before touching a policy, because that is where the fix actually is.

By working out why visits fail, which is normally three things: the wrong skill was dispatched, the part was not on the van, or the person arriving did not know what happened on the last visit. Each has a different fix, and each is answerable from your own history if the data is there. We chase first-time fix ahead of raw efficiency because a failed visit costs a second dispatch, a second travel leg and a customer who has now taken two days off work.

Two things. It asks for more than the job needs, so a five-minute repair carries a fifteen-minute form. And it fails where the work happens: basements, lift shafts, plant rooms, rural sites. Offline behaviour is a design decision made up front, not something discovered during the pilot. We design the mobile flow around gloved hands, bad light and no signal, then cut the data capture back to what someone downstream genuinely uses.

Yes, and it is worth designing deliberately rather than bolting on. Contractors need to receive work, see enough to do it, and provide evidence and cost back, without being given the run of your org. We model that access explicitly, and we treat their capacity as real capacity in the schedule rather than as an overflow tray that dispatch remembers to use when everything is already late.

Maintenance plans generate work from the asset, which is the right model, but the practical failure is bunching: every asset commissioned in the same quarter comes due in the same week, and the fleet cannot absorb it. We spread generation against real capacity. Asset history matters just as much, because a technician who can see the last three visits fixes things a technician arriving cold does not.

By making it explicit rather than letting the policy decide it silently. Optimisation objectives are weighted, and those weights encode a commercial choice: minimising travel will sometimes miss a window, and guaranteeing every window will sometimes cost overtime. We put the current trade-off in front of the people who own the P&L and the customer promise, because that decision belongs to them and not to a scheduling policy.