Skip to main content

Costliest clients

The Costliest clients report ranks your sites by how much they actually cost your control room to run — in Rand — so the expensive minority is obvious. In most operations a small share of sites drives the large majority of effort (the classic 80/20), and this report makes that concrete and actionable.

Open it under Review Queue → Costliest clients, and pick a window (30 / 90 / 180 days).

What it costs​

Each site's cost is built from four centres, priced by tunable rates:

  • Operator — the controller time spent handling that site's events (and calls), at a rate derived from the controller monthly cost.
  • Call — call attempts, costed as operator time by default (or a flat per-call rand if you set one).
  • Reaction — reaction-unit labour: the measured time on the job (dispatch start → complete) when available, otherwise a default estimate.
  • Transport — distance travelled (from the dispatch and site locations) at a per-km rate, round-trip by default.

Only events that actually reached an operator are counted — raw signals that auto-resolve are not.

Data quality

The report clamps implausible values (e.g. a dispatch left open for days, or a bad GPS coordinate producing a thousand-kilometre trip) so one bad record can't dominate the ranking. Sites that surface with extreme numbers are often worth investigating for data-quality reasons too — a never-closed dispatch or an incorrect site coordinate.

Reading the report​

  • Four stat cards summarise the window: total modelled cost, active sites, how many sites make up 80% of the cost, and the costliest site with its amount.
  • The table lists each site by rank with its event / call / dispatch counts, the four cost centres, the total, and its share of the whole (as a bar). The rows that together make up the first 80% of cost are highlighted with a warm tint — those are the clients to focus on.
  • Each site name links straight through to that site's page, so acting on a finding is one click.

Automatic cost alerts (Review Queue)​

Beyond the report, a weekly job watches each control room's dispatch behaviour and raises review-queue items when a site stands out — adaptively, against that control room's own average (not a fixed number):

  • High dispatch cost — a site dispatching far more than its peers, weighted by wasteful (cancelled / false) runs.
  • Vehicles rolled without contact — a site where vehicles are sent without reaching the customer or keyholder (a contactability / keyholder problem).
  • Keyholder rarely answers — a specific listed contact who repeatedly doesn't answer (and whether they're the only contact), so you can add a reachable backup.

Each item carries the numbers and a concrete suggestion, and appears in the Review Queue → Queue tab where it can be snoozed, resolved, or actioned. These are governance prompts — they never change anything automatically.

Tuning the rates​

The rates live on Settings → Cost model (under Operations). There is a global default, and each control room can set its own override (and Reset to default to drop it). Anyone who can manage the control room may edit them — these are your own operating costs, so your managers set them, not CleverCam. Everyone else sees the current values read-only. Saving only ever writes this control room's override; the global default behind it is CleverCam's.

Tunable values include the controller monthly cost and productive hours, seconds per event and per call, the reaction default minutes and officer rate, the transport per-km rate, the round-trip and measured-reaction toggles, the outlier clamps, and an optional flat per-call cost.

note

The rates ship with sensible placeholders — set the reaction-officer rate and transport per-km rate to your real figures for accurate rands. The ranking of sites is robust to rate changes; the absolute amounts scale with them.