Start by separating the cost ledgers
A utility repair invoice records labour, materials, contractors, logistics, and related restoration work within an organization's accounting boundary. SAIDI records customer interruption duration. Insured catastrophe reporting covers insured property and casualty losses under its own event and dollar definitions. A capital plan describes future spending and targets. Business interruption, health, safety, food spoilage, lost production, and essential-service consequences may fall partly or entirely outside those measures. The figures can all be relevant without being additive. Sources: Hydro-Québec (opens in a new tab) Hydro-Québec (opens in a new tab) Hydro-Québec (opens in a new tab) Hydro-Québec (opens in a new tab) Insurance Bureau of Canada (opens in a new tab)
The error in a mixed-scale chart is not only mathematical. Placing an event repair cost, an annual reliability indicator, a decade-long capital plan, and nationwide insured loss on one normalized bar scale invites readers to infer a comparison the sources do not support. A scope-safe presentation identifies organization, geography, period, metric, nominal or adjusted dollars, and whether the number is an observed outcome, a plan, or a target. This article therefore uses separate panels and a table explaining why the numbers cannot be totaled. Sources: Hydro-Québec (opens in a new tab) Hydro-Québec (opens in a new tab) Insurance Bureau of Canada (opens in a new tab)
What reactive grid management means
Reactive management begins when the organization has confirmed impact: an outage appears, customer calls or telemetry arrive, damage is assessed, crews are assigned, materials are found, access is established, repairs are completed, and service is restored. This work will always be necessary because no forecast can prevent every tree failure, accident, equipment defect, or weather surprise. The problem is not restoration itself; it is entering the event with avoidably little shared context about where pressure may build and which preparatory decisions deserve review.
Earlier intelligence might change monitoring intensity, on-call posture, inspection priority, material checks, mutual-aid coordination, customer communications, or essential-service watchlists. It can also create cost through unnecessary staging, false alarms, data integration, analyst review, and maintenance of the forecasting system. A business case must count both sides. The relevant comparison is not a dramatic event total versus software cost; it is the outcome of a defined decision under a measured baseline and intervention. Source: National Institute of Standards and Technology (opens in a new tab)
Evidence: Conceptual Illustration
Where earlier evidence might change a decision
Show decision opportunities without assigning an unmeasured saving.
-
Before impact
Review
Forecast, exposure, readiness, essential services
-
Approaching event
Prepare
Monitoring, materials, communications, coordination
-
Confirmed outage
Respond
Safety, assessment, switching, repair, restoration
-
After event
Measure
Decisions, costs, outcomes, misses, false alarms
Show decision opportunities without assigning an unmeasured saving.
Reliability measures are not dollar measures
SAIDI commonly summarizes the average total interruption duration experienced per customer over a reporting period. SAIFI summarizes interruption frequency, and CAIDI relates interruption duration to customers interrupted. Their precise treatment depends on the reporting methodology, including momentary interruptions, planned events, exclusions, and major-event adjustments. A comparison must use the same utility scope and metric basis. Converting minutes directly into dollars requires additional customer and interruption-cost assumptions that the reliability indicator does not contain. Sources: Hydro-Québec (opens in a new tab) Hydro-Québec (opens in a new tab) Hydro-Québec (opens in a new tab)
Hydro-Québec reported unadjusted SAIDI of 1,072 minutes per customer for 2023 and said five major events contributed 80 percent of overall SAIDI. Its 2024 distribution reporting lists 436 unadjusted minutes. This consistent utility-and-measure comparison shows how strongly event mix can shape annual customer experience. It does not prove why the value changed, isolate the effect of one program, or establish a Canada-wide trend. Adjusted and unadjusted measures should never be swapped without explanation. Sources: Hydro-Québec (opens in a new tab) Hydro-Québec (opens in a new tab) Hydro-Québec (opens in a new tab)
Evidence: Observed Public Data
Hydro-Québec unadjusted SAIDI
Compare one reliability measure for one utility while preserving the major-event context.
Scroll horizontally or use the arrow keys to compare every column.
| Reporting year | Unadjusted SAIDI | Context |
|---|---|---|
| 2023 | 1,072 minutes | Five major events contributed 80% of overall SAIDI |
| 2024 | 436 minutes | Utility filing; do not infer one-program causality |
Minutes per customer, unadjusted. This two-year comparison does not establish causality or a national trend.
Figure sources: Hydro-Québec (2025) (opens in a new tab) Hydro-Québec (2024) (opens in a new tab) Hydro-Québec (2024) (opens in a new tab)
Two Quebec events show the scale—and the boundaries—of restoration work
Hydro-Québec's May 2022 derecho recap reported 11,254 outages, 554,649 customers interrupted at 8 p.m., and approximately $70 million of work. The same source described extensive pole, transformer, conductor, and vegetation work across the event. Those numbers are observed outcomes within Hydro-Québec's reported Quebec scope. They are not a prediction result, not a national outage total, and not a complete estimate of business or community losses. Source: Hydro-Québec (opens in a new tab)
For the April 2023 ice storm, Hydro-Québec reported 9,669 outages, 1.31 million customers affected in total, 1.125 million simultaneously, and approximately $50 million in restoration cost. The outage count corrects the legacy article's inaccurate 'more than 2,000' wording. Even these two utility event costs should not be casually added or ranked as a recurring annual burden: the storms, accounting periods, damage patterns, and reporting boundaries differ. Source: Hydro-Québec (opens in a new tab)
Evidence: Observed Public Data
Two events, two reported scopes
Present observed Hydro-Québec event facts without adding them into a national total.
- May 2022 derecho 11,254 outages
- 554,649 customers interrupted at 8 p.m.; about $70M of work
- April 2023 ice storm 9,669 outages
- 1.31M customers in total; about $50M restoration cost
Nominal reported event figures; each card retains its utility, geography, and event scope.
Figure sources: Hydro-Québec (2022-06-14) (opens in a new tab) Hydro-Québec (2023-05-15) (opens in a new tab)
Customer interruption and community consequences extend beyond repairs
An interruption can halt production, close retail operations, spoil inventory, disrupt remote work, disable building systems, and force households or organizations to use backup power. The cost varies sharply with customer type, outage timing and duration, advance notice, season, backup capability, and whether dependent communications or transport are also affected. A single value-of-lost-load assumption cannot be applied across every Canadian customer without explaining the survey, sector, dollar year, and scenario behind it. Source: Government of Canada-led assessment (opens in a new tab)
Essential services add consequences that are not well represented by a simple commercial loss estimate. Hospitals may have generators but still depend on fuel, water, telecom, access, staff, and suppliers. Water systems can require pumping; telecom requires power; payment and fuel systems can depend on communications. Duration and cascading dependencies matter as much as geographic proximity. Public discussion should remain at a safe level and avoid exposing operational vulnerabilities merely to make a cost story vivid. Source: Government of Canada-led assessment (opens in a new tab)
Insured severe-weather loss is context—not grid cost
Insurance Bureau of Canada reported $8.5 billion in Canadian severe-weather insured losses for 2024, based on CatIQ data. That national figure spans covered property and casualty catastrophe damage. It is not a utility restoration bill, a measure of outage duration, or an estimate of electricity-related economic loss. It belongs in a separate context panel with the source, year, geography, insurance scope, and nominal-dollar basis visible. Source: Insurance Bureau of Canada (opens in a new tab)
Later reporting also illustrates why historical dollar comparisons need a stated inflation basis and revision date. Catastrophe totals can be updated as claims develop, and one article may quote nominal event-year dollars while another uses inflation-adjusted dollars. The safe practice is to cite the specific release and preserve its wording rather than silently harmonize amounts. These figures show the broader financial environment in which grid resilience matters; they do not quantify the portion an outage tool could change. Source: Insurance Bureau of Canada (opens in a new tab)
Evidence: Observed Public Data
Why prominent numbers cannot be totaled
Keep unlike measures in separate rows with their own units, geography, period, and status.
Scroll horizontally or use the arrow keys to compare every column.
| Figure | What it measures | Why it is not additive |
|---|---|---|
| About $70M | Hydro-Québec work after the May 2022 derecho | One utility and event |
| 436 minutes | Hydro-Québec 2024 unadjusted SAIDI | Reliability duration, not dollars |
| $45–50B | Planned investment through 2035 | Multi-year plan, not realized event cost |
| $8.5B | 2024 Canada insured severe-weather loss | National insured catastrophe scope |
Keep unlike measures in separate rows with their own units, geography, period, and status.
Figure sources: Hydro-Québec (2022-06-14) (opens in a new tab) Hydro-Québec (2025) (opens in a new tab) Hydro-Québec (2023-05-15) (opens in a new tab) Hydro-Québec (2023-11) (opens in a new tab) Insurance Bureau of Canada (2025-01-13) (opens in a new tab)
Investment plans and targets are not achieved savings
Hydro-Québec's Action Plan 2035 describes planned investment of $45–50 billion through 2035 and a goal to reduce outages by 35 percent over seven to ten years. Those figures communicate the intended scale and direction of work. They are not an observed cost of one outage, not a completed investment total, and not evidence that the reduction target has been achieved. Time horizon and status must travel with the number wherever it appears. Source: Hydro-Québec (opens in a new tab)
Vegetation-management announcements similarly describe planned activity and goals. Hardening, automation, maintenance, pruning, inspection, and data systems can all contribute to resilience, often together. A later reliability improvement cannot be assigned to one component without an evaluation design that addresses weather severity, asset work, customer growth, reporting changes, and other confounders. Planning documents are evidence of commitment and intended action, not causal outcome evidence. Source: Hydro-Québec (opens in a new tab)
What earlier intelligence could change—and what remains unproven
A useful risk system may provide lead time to review crews, materials, vegetation corridors, essential-service dependencies, communications, or field inspection priorities. Those are intermediate decisions, not guaranteed avoided outages. The value may appear as faster recognition, shorter travel, fewer unnecessary inspections, better use of standby time, improved coverage of critical locations, or more accurate briefings. Each outcome needs a definition and a data-collection plan before the event. Source: National Institute of Standards and Technology (opens in a new tab)
This article has no public evidence that GeoGridIQ reduces restoration cost, interruption minutes, customers affected, or insured loss by a fixed amount or percentage. It also has no deployment study showing that a recommendation changed an operator decision. The appropriate claim is narrower: prediction and geospatial context could support earlier review, and their value can be tested. Presenting that hypothesis honestly makes a later evaluation stronger because the baseline and promised outcome are not rewritten after results are known. Source: National Institute of Standards and Technology (opens in a new tab)
A measurement framework for proving value
Begin with a specific decision, such as whether to stage a crew, inspect a corridor, pre-position material, or add an essential-service watch. Record the existing process and cost, the eligible event population, the forecast issue time, action threshold, people involved, and expected outcome. Then retain predictions and decisions prospectively. Compare lead time, false and missed alerts, action frequency, travel, labour, restoration milestones, customer interruption, and safety-relevant constraints with appropriate controls. Source: National Institute of Standards and Technology (opens in a new tab)
Count the full intervention cost: data licensing or collection, integration, compute, monitoring, analyst and operator time, training, governance, and unnecessary actions. Segment results by region, horizon, hazard, provider coverage, and data state. Report uncertainty and adverse outcomes. A dashboard metric is not a causal study, and before-versus-after improvement during different storm seasons is not enough by itself. This scorecard turns a broad business case into a falsifiable operational evaluation. Source: National Institute of Standards and Technology (opens in a new tab)
Evidence: Conceptual Illustration
How to test the value of predictive intelligence
Define a decision and baseline, retain the intervention, measure outcomes and full costs, then report uncertainty.
-
1
Decision
What action might the information change?
-
2
Baseline
How is it done now, and at what cost?
-
3
Prospective record
Store forecasts, review, actions, and context
-
4
Outcomes
Lead time, travel, duration, coverage, burden
-
5
Full accounting
Include integration, review, errors, and uncertainty
Define a decision and baseline, retain the intervention, measure outcomes and full costs, then report uncertainty.
Scope and safeguards
Limitations and responsible use
- Most quantitative utility evidence here is scoped to Hydro-Québec rather than all Canadian utilities.
- Reliability definitions, exclusions, and major-event treatment must match before comparison.
- Dollar figures use different years, accounting boundaries, geographies, and statuses.
- Planned investment and reduction goals are not achieved outcomes.
- No causal or return-on-investment evaluation of GeoGridIQ is presented.
Frequently asked questions
Questions this article answers
What does a power outage cost a utility?
It depends on the event and accounting boundary. Labour, contractors, materials, logistics, and restoration work are different from customer and community losses.
What are SAIDI and SAIFI?
SAIDI summarizes interruption duration per customer; SAIFI summarizes interruption frequency. Exact reporting and major-event treatment must accompany comparisons.
Are insured weather losses the same as grid outage costs?
No. Insured catastrophe losses cover insured property and casualty damage across an event; they are not utility restoration or outage-cost totals.
Can outage prediction eliminate restoration costs?
No. It may support earlier decisions, but faults and restoration work remain, and the tool itself creates costs and false-alarm trade-offs.
How should a utility measure predictive value?
Define a decision, baseline, eligible events, intervention, full costs, and prospective outcomes before inspecting results.
Evidence register
Sources
Sources were reviewed on . Mutable sources are rechecked on the article review schedule.
-
D2 Hydro-Québec. A look at the outages caused by the May 21 derecho (opens in a new tab). 2022-06-14.
-
R1 Hydro-Québec. 2024 Distribution Activities (opens in a new tab). 2025.
-
R2 Hydro-Québec. Sustainability Report 2023 (opens in a new tab). 2024.
-
R3 Hydro-Québec. 2023 Distribution Activities (opens in a new tab). 2024.
-
R5 Hydro-Québec. April 2023 ice-storm recap and resilience actions (opens in a new tab). 2023-05-15.
-
R6 Hydro-Québec. Action Plan 2035 (opens in a new tab). 2023-11.
-
R7 Hydro-Québec. Improving electricity service quality (opens in a new tab). 2024-02-27.
-
E1 Insurance Bureau of Canada. 2024 severe-weather insured losses (opens in a new tab). 2025-01-13.
-
E2 Insurance Bureau of Canada. 2025 insured losses and revised historical comparison (opens in a new tab). 2026-01-20.
-
A1 National Institute of Standards and Technology. AI Risk Management Framework 1.0 (opens in a new tab). 2023-01-26.
-
C2 Government of Canada-led assessment. Canada in a Changing Climate — National Issues, Chapter 2 (opens in a new tab). 2021.