Performance-based regulation reporting with every reward and penalty traced to its source records
Every reward and penalty traced to its source records, and explained to the regulator.
See one case, screen by screen ↓Under performance-based regulation every standing metric — outage minutes and frequency, calls answered, complaints, estimated bills, solar and battery connection times, energy savings, lost water — carries a target, a deadband, a dollar rate and a cap set in the tariff. The data behind them sits in ten systems: the outage management system, the smart-meter head-end, the network model, billing, the call platform, the interconnection portal, programme tracking, water balance, the lab system and the ledger. The regulatory team pulls a year of it into spreadsheets and rebuilds each calculation by hand.
The methods live in personal workbooks and drift from year to year. The explanation for the regulator is written last, in Word, and cites nothing — so when staff ask where a number came from, the team reverse-engineers its own workbooks. And the incentive rides on decimals: one storm day assigned to the wrong date, one outage entered twice or a sustained outage logged as momentary moves the reliability index, and with it a reward or a penalty.
Estimated split for one performance year's scorecard and narrative, in working days, by hand and with the solution.
Six specialist agents collect, calculate, reconcile, price and explain every metric; the metric owner decides each correction and the Director signs.
Lumina Utilities, electricity and water, serves 612,400 electric customers. Its performance year 2026 report is due to the Halvern Utilities Commission on Dec 1, under Docket 26-PBR-0417 and tariff Schedule PBR-1. On Oct 13, 2026, the morning run leaves reliability in penalty. Here is what happens next, screen by screen, in the working solution.
The daily 06:00 run starts. The source collector takes the extracts at the Oct 10, 06:00 cut-off from ten systems — 2.94M records — and checks each month of each system for completeness. Six items are flagged for their owners, none of which blocks the calculation: storm interruptions restored after midnight in August, 1.3 % of water reads still estimated at data freeze, a 2-day call platform outage backfilled from the carrier's call records.
The source collector never fills a gap with an estimate; it reports it to the owner.
Filing is due Dec 1 — 49 days — with data freeze on Nov 3. The run computed all 13 metrics, 10 with money at stake: net incentive −$2.35M, rewards +$1.75M against penalties −$4.10M. Outage minutes per customer (SAIDI, excluding major event days) came out at 121.4 min against a 118.0 target and a ±2.0 deadband: 1.4 minutes past the penalty edge, at $1.5M a minute, −$2.10M.
“3 corrections waiting on E1, E2 · +$4.75M if accepted”
Every bar runs from the penalty cap on the left, through the grey deadband around the target, to the reward cap on the right. A dashed marker shows where SAIDI and SAIFI would land if the open corrections were accepted. Tomás switches to With proposed corrections: SAIDI 113.2 min, SAIFI 1.01, net +$4.75M — marked “Preview — not decided”.
“Preview: E1 and E2 with the 3 proposed corrections — nothing is changed until Grace Liu decides”
Tomás opens E1. Four sources feed it: 14,812 outage records, 2,310,450 smart-meter power-off and power-on events, the major event day register (5 days, T_MED 6.84) and 612,400 customers served. Four tariff steps follow — keep sustained interruptions (§3.1), remove duplicates (§3.4), assign each interruption to the day it begins (§3.2), exclude major event days (§3.3) — down to 74,345,360 customer minutes, 121.4 min and −$2.10M under §4.1.
Click any step to see its records — every step names its tariff section and its record count.
The reconciler cross-checked the outage records against the meter events and the major event day register, and proposes three corrections, each with both records side by side. 1,214 interruptions began on Aug 23, a major event day, were restored after midnight and were counted on Aug 24 (−7.5 min). Feeder F-22 outage OMS-26-118842 was entered again as OMS-26-118867 on a crew re-dispatch (−0.8 min). Three recloser events on F-07, F-31 and F-44, logged as momentary, show 7,870 meters off for 7.2 to 8.4 minutes — sustained (+0.1 min).
“+0.1 min · against Lumina” — the reconciler reports corrections that raise a penalty, not only those that lower it.
Grace owns E1 to E3. She accepts all three corrections in one step, signing as the metric owner, and her reason goes into the audit trail. In the same step she marks the E1–E3 explanations reviewed. No correction changes a metric until its owner accepts it.
Reason: “Checked against the Aug 23 storm log and the F-22 crew tickets; matches Schedule PBR-1 §3.1–3.4.”
SAIDI is now 113.2 min — 69,323,680 customer minutes over 612,400 customers — 2.8 minutes beyond the reward edge, +$4.20M. The paragraph for the Commission says so, adds a sentence on the three corrections with each one cited to its step, and adds a method-change note: PY2025 assigned interruptions to the day they were restored; restated on the PY2026 basis, PY2025 SAIDI is 119.6 min against 124.1 filed.
“Method change · disclosed” — the narrative writer can never remove a method-change disclosure.
E1 moves to +$4.20M, and E2 from $0 inside its deadband to +$0.80M at 1.01. Net incentive +$4.75M: rewards +$6.75M, penalties −$2.00M. The penalties that stand are traced as well — calls answered within 30 seconds at 76.4 % (−$0.40M), small solar and battery applications approved in 15 business days at 82.0 % (−$1.00M), lost water at 14.4 % (−$0.60M).
“Reliability moved from penalty into reward: E1 113.2 min, +$4.20M”
Commission staff sent four questions ahead of the technical conference on Nov 5. HUC-DR-3-12, due Oct 27, asks for the workpapers behind SAIDI and SAIFI and the interruptions removed on each major event day. The draft answer names the five days — Dec 11, Jan 17, Mar 4, Jun 28 and Aug 23 — and the 2,388 interruptions excluded, and attaches workpaper tabs 1–6, Appendix C and the E1 trace export. It goes to Elena Marsh for review.
“Staff will see the attachments and the answer; the trace stays internal unless it is attached.”
The filing assembler checks 8 of 8 items ready: the scorecard table, 13 workbooks with every figure linked to its records, the major event days appendix, the discussion of results with 13 of 13 sections reviewed by their owners, two method-change disclosures and the data dictionary. Elena signs at 10:24. Marcus Hale, the CFO, attests the net incentive of +$4.75M at 10:31, and the report is filed under Docket 26-PBR-0417.
The filing assembler never submits — the Director signs and the CFO attests.
The same performance year, seen by the five people who own it — what reporting season looked like, and what it looks like now.
Six specialist agents collect, calculate, reconcile, price and explain every metric; the metric owner decides each correction and the Director signs.
Pulls the performance year's data from the ten source systems, counts the records and checks every month is complete, flagging any month that is missing, partial or late.
Computes each metric exactly as the tariff defines it and records every step — the rule and its section, the records in and out, and a sample — as the metric's trace.
Cross-checks the sources against each other — outage records against smart meters, duplicates, day assignment, customer counts against billing — and proposes corrections with evidence.
Applies each metric's target, deadband, rate and cap and prices the reward or penalty and the net, again whenever a metric value changes.
Drafts the explanation for each metric and the answers to staff questions; every sentence cites the trace and every number is bound to it.
Builds the filing — scorecard, workpapers, narrative, disclosures, attestation — checks every number in the narrative against the scorecard and the workpapers, and checks it against the Commission's filing requirements.
Accept, sign and attest. The agents propose; a named person decides.
Ask about any metric, its calculation or its explanation — or say what to change. Every answer points to the step of the trace it came from.
Why would reliability move from penalty to reward?
Three corrections, all found by tracing SAIDI to its records. 1,214 storm interruptions that began on Aug 23, a major event day, were counted on Aug 24; the tariff assigns them to the day they begin. One F-22 outage was recorded twice, and three outages logged as momentary lasted 7–8 minutes on the meters, which adds 0.1 minutes. Net: 121.4 → 113.2 minutes, from −$2.10M to +$4.20M.
Which metrics are closest to a deadband edge?
Four are within one rate step of an edge: E2 Outages per customer, 1.01 with the open corrections — 0.01 past the reward edge, worth $0.8M; W1 Lost water at 14.4 %, 0.1 past the penalty edge (−$0.6M); P2 Peak demand reduction at 97 MW, 3 MW from the reward edge; and C2 Complaints at 0.36, 0.01 past the reward edge.
Draft the answer to staff on method changes
Drafted HUC-DR-3-14 from the method change register. Two changes from PY2025: interruptions assigned to the day they begin (E1, E2), and the water balance v6 counting flushing water as authorised use (W1). PY2025 restated on the PY2026 basis: SAIDI 119.6 (filed 124.1). It is in Data requests for Elena Marsh to review.
Hold any explanation that has no owner review
Done. New filing rule: an explanation goes into the filing only after its metric owner marks it reviewed. Today that holds 3 of 13 sections, E1, E2 and E3, all waiting for Grace Liu. You can change it in Settings.
13 screens from the working solution, on its sample data. Pick one to see it large.
Every metric rule comes from your tariff, and the reconciliation checks, approvers and clocks are settings. Changes need the Director's approval and apply from the next run.
| Setting | Default | Choose from |
|---|---|---|
| Metric rules | 13 metrics · Schedule PBR-1 §4–§5 | Per metric: better when lower, higher or reporting only, target, deadband, rate, cap, owner and tariff section |
| Major event days | IEEE 1366, 2.5β | Base period: daily SAIDI, five years to Sep 30, 2025 · threshold T_MED 6.84 min |
| An interruption belongs to | The day it begins (§3.2) | Momentary below 5 min |
| Outage records vs smart meters | On | On · Off — flags a duration that differs by more than 2 min |
| Duplicates | On | On · Off — same device, windows overlap more than 30 min |
| Customers served vs billing | On | On · Off — flags a difference above 0.5 %; complaints are matched to the Commission's log |
| Report corrections that raise a penalty | Always on | On — the reconciler reports corrections in both directions, those that raise a penalty as well as those that lower it |
| Approvals and clocks | Correction: metric owner · filing: Elena Marsh · net incentive: Marcus Hale | Data freeze Nov 3 · filing due Dec 1, warn 14 days before |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “target” = the design goal, measured in the live solution · “estimated” = our estimate of where a scorecard's working days go · “typical” = published figures (RMI, Shining a light on utility performance in Hawaii, 2021 — almost 40 metrics and scorecards; Illinois multi-year rate plan order, 2023 — seven metrics worth ±24 basis points of ROE). Context: IEEE 1366 (major event days); OEB electricity distributor scorecard; Hawaiian Electric performance scorecards and metrics. Lumina Utilities, the Halvern Utilities Commission, Docket 26-PBR-0417, Schedule PBR-1 and the people and records named on this page are fictional — characters and sample data in the working solution.
Under performance-based regulation a utility earns a reward or pays a penalty on a set of standing metrics, each with a target, deadband, rate and cap in its tariff, and reports the results to its regulator every year with an explanation. Performance-Based Regulation Reporting computes every metric from the source systems, traces each reward and penalty to its records, and drafts the cited explanation; people decide, sign and attest.
Six agents collect the year's data from ten systems, compute each metric exactly as the tariff defines it with every step recorded, cross-check the sources against each other, price each reward or penalty, draft the explanation for the regulator and assemble the filing. The scorecard shows every metric on a bar from penalty to reward, and each one opens its trace.
It follows the method set in Settings — in the sample, IEEE 1366 with the 2.5 beta method on five years of daily SAIDI, giving a threshold of 6.84 minutes. Interruptions are kept if sustained (over 5 minutes), de-duplicated, assigned to the day they begin and excluded on major event days, and each step shows its record count.
No. Reporting corrections that raise a penalty is always on. In the sample case the reconciler found two corrections that lowered SAIDI and one — three sustained outages logged as momentary — that raised it by 0.1 minutes against the utility.
Yes. Staff questions sit in Data requests with their due dates, and the narrative writer drafts each answer from the same trace that produced the number, with the workpapers and appendices attached. The Director of Regulatory Affairs reviews and signs every answer.
Yes. The agents have read-only access to the source systems; they calculate, propose and draft but never accept, sign or file. Each correction needs its metric owner, the filing needs the Director's signature and the net incentive the CFO's attestation.
Yes. Each metric's direction, target, deadband, rate, cap, owner and tariff section, the major event day method, the momentary threshold, the reconciliation checks, the approvers and the filing clocks are settings. Rule changes need the Director's approval and apply from the next run.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your tariff, the commission's filing requirements and last year's scorecard and workpapers — and it goes live once every quality gate has passed. We will walk you through it on your own performance year first.
Every data request answered on time, with one number for one thing.
1-min demoRegulatory Precedent SearchEvery filing position backed by how regulators actually ruled, cited and trended.
1-min demoRegulatory Commitment TrackingEvery directive owned and dated within a day of the order, and closed with its proof.
1-min demoWe'll run Performance-Based Regulation Reporting on last year's metrics, under your own tariff.