Operations observatory
The operators’ own figures, month by month.
Monthly operational figures from Sparkoper deployments, published with each operator’s agreement. They show what changed while the platform and the support around it were in place — not a claim that the software alone produced the change. Two deployments are on this page today; others will join as their operators agree.
Method. Every figure is recomputed from the raw export at each update, never typed in by hand. An indicator is published as a monthly curve only when each month carries enough measurements for the monthly figure to mean something. Where it does not, the indicator is given as a single figure over the whole period, and labelled as one.
EPAL-EP · Angola · water utilityData as of 12 November 2026Tracked since June 2026Next update December 2026
Scope. Maintenance of treatment equipment at one plant. Planned work orders and corrective requests, one maintenance team. The scope will widen as further sites are added, and raw volumes will rise with it.
Planned maintenance done on time
Share of planned work orders completed on or before the date they were planned for, counted in the month the work was due. Measured on the 191 of 359 planned orders that carry both a planned date and a completion date.
MeasuredProjection
Each point is a monthly figure. The dashed segment prolongs the recent trend until the next monthly update, when it is replaced by the measured value.
| Month | Value |
|---|---|
| June 2026 | 12% |
| July 2026 | 44% |
| August 2026 | 54% |
| September 2026 | 77% |
| October 2026 | 83% |
1.7days
Median time to close a corrective order
Median time from a corrective request being raised to its closure, over the whole period. Too few orders per month — between two and fifteen — to carry a monthly curve.
Over 92 orders.
68%
Planned orders completed
Share of all planned work orders that have been completed. The counterpart of the curve above: punctuality improves, and a backlog remains.
Over 359 orders.
AGRP · Brazil · urban servicesData as of 12 November 2026Tracked since November 2025Next update December 2026
Scope. Operational requests across Reserva do Paiva. A video control room detects most incidents and closes them on sight; what follows counts only the requests it escalates to a named field agent.
2.9days
Median time to close an escalated request
Median time from detection to closure, counting only requests raised by the control room and resolved by a different, named agent. Between five and twenty-seven such requests a month, too few for a monthly curve.
Over 159 requests.
82%
Escalated requests closed within 30 days
Share of escalated requests closed within thirty days of detection, over the whole period.
Over 159 requests.
How these figures are made
- 01Computed from monthly aggregates produced inside the operator’s own platform. No individual record, no name and no address leaves it.
- 02Each figure is a count, a median or a share, computed without a model. The scope it covers is stated above it.
- 03The current month is a projection until it is measured. It is drawn dashed, with a range, and replaced by the measured value at the next update.
- 04A deployment appears here in full. We choose when it joins the page, never which of its indicators are shown — and an indicator published stays published.
- 05Published with the agreement of the operator, who is named above. What these figures establish is what changed, and when; turning that into attribution is the reader’s judgement, not our claim.
Want these figures for your own operations?
We set up the same indicators on your data, inside your systems.