Preview · sample

outage forecasting for restoration contractors

Fewer dead miles. More billable storms.

A good year or a bad one comes down to where you send the trucks.

You have a finite number of crews, a finite number of trucks, and only so much windshield time. This 1–5 day outage forecast ranks where restoration work is actually likely to land — so you position for the work instead of guessing at it. It's a forecast of where work is likely, not a probability or a customer count.

The outlook

Counties ranked by outage-work priority.

Lead time
Counties & regions ranked by outage priority — day 1 tap a row for counties

The same forecast, on a live weather map.

Basemap © OpenStreetMap, © CARTO · Wind: Open-Meteo · Rain imagery: RainViewer
Extreme High Moderate marker color = priority tier (not outage size)

How to read it

Priority tiers and historical rates.

Each county gets a priority tier from the forecast. The percentages are how often counties at that rank actually had outages in the validation data — historical fact, not a forecast probability. The signal weakens with lead time, so detail coarsens from counties (near-term) to regions (far out).

Extreme
~36% historical
outage rate in validation · held most of total restoration hours
High
~20% historical
elevated · worth positioning toward near-term
Moderate
~12% historical
a noticeable lean above background
Normal
background
no elevated signal
🌀

It follows the weather as it moves.

The forecast is driven by the evolving path of wind and rain. The weather model tracks each system as it develops and moves across the map — day 1, day 2, day 3 — and this model reads that forecast county by county, so the priority shifts with the storm. As a system moves toward a region, the counties in its way light up ahead of time; as it passes, they cool down. You're not seeing a static snapshot — you're seeing where the work is headed, because the forecast underneath it is tracking exactly that.

10% of storms carry 88% of the work.

That's the number that should change how you operate. Restoration work isn't spread evenly — in validation, the top 10% of storm-days held ~88% of all restoration hours. Chasing every little event burns crews and miles on work that was never there. The win is the opposite of reactive: know which handful of storms actually matter, and put everything into those. This tells you which ones.

What it cannot do (stated plainly)

  • It does not predict how many customers lose power, or for how long. It ranks where work is likely to concentrate.
  • No peak-timing window or hour-by-hour forecast — the signal is daily.
  • The tier is a relative priority, not a calibrated percent chance.
  • Validated on warm-season storm/convective events in covered regions; cold-season is in development and not shown.
  • Validated on held-out storms within past years — a brand-new season is reasonable to expect but not yet proven.

Know when your region lights up.

Register and we'll notify you when the forecast flags elevated outage work in your region — so you're positioning while everyone else is still watching the news. No spam, unsubscribe anytime.