← DOWNTIME RISK ASSESSMENT

Methodology

Every dollar figure on the Downtime Risk Assessment is an estimate modeled from public infrastructure data and industry benchmarks — it is not a quote, an audit, or a measurement of your business. Every factual claim about an address traces to a public dataset row with a source URL and retrieval date. Benchmarks sit at the conservative (low) end of published ranges. Benchmark set version: 2026.07-p1.

The invoice model

Annual cost = per-outage-hour exposure × expected outage hours/year at your address, itemized as:

001 LOST SALES = effective outage hrs × hourly revenue × locations 002 IDLE LABOR = effective outage hrs × employees × loaded $/hr × locations 003 FAILED CARDS = lost sales × 0.09 004 EMERGENCY IT = max(2, regional events ÷ 3) × $260 × locations 005 CUSTOMER CHURN = affected customers × churn% × avg ticket × visits/yr (per vertical)

Effective outage hours scale the address’s expected annual outage hours to your open hours only (outages outside business hours don’t bill you).

Line 003 — degraded windows. Failed card transactions are counted in addition to lost sales because they occur in degraded-but-not-down windows (partial packet loss, DNS instability, modem flap) that are distinct from the full-outage hours in line 001. During these windows the doors are open and customers are buying — but card authorization fails. The 9% factor is a modeled constant at the low end of published payment-failure impact during connectivity degradation.

Line 004 — emergency IT. $260 is a modeled blended cost per emergency response (typical published SMB emergency IT rates run $125–$250/hour plus trip fees, with a 1–2 hour median engagement). Incident count is modeled as the outage-event baseline ÷ 3, floored at 2/year.

Line 005 — churn. Visit frequency is vertical-specific (a dental patient returns ~2×/year; a convenience-store regular ~35×), so the annual value of a lost customer differs by an order of magnitude across verticals. Churn percentages sit at the low end of published walk-away/no-return rates.

Per-vertical benchmarks

VERTICALREV/HRTICKETTX/HRLABOR/HRCHURNVISITS/YR
Restaurant / QSR$680$1819$216%15
Retail$560$427$195%5
Grocery / C-Store$1040$3117$204%40
Fuel Station$1220$2426$193%35
Medical / Dental$900$1902.4$348%2
Salon / Services$320$652.5$227%6
Auto Repair$780$3101.3$285%1.5
Pharmacy$620$605$345%12
Hotel / Lodging$500$1403$225%1.5
Other$600$456$245%6

The outage-risk grade

Four factors compose a 0–100 score, then a letter grade (A ≥ 85, B ≥ 72, C ≥ 60, D ≥ 45, F below):

REDUNDANCY 40% distinct wired ISPs at the location (FCC BDC) OUTAGE HISTORY 25% 12-mo county disruption index (DOE EAGLE-I + EIA-861) INFRASTRUCTURE AGE 20% copper retirement/discontinuance status ENVIRONMENTAL EXPOSURE 15% FEMA National Risk Index (hurricane, ice, wildfire, tornado)

This weighting is deliberately redundancy-first: a single wired provider is the dominant, controllable outage risk for a fixed address. Grades skew tough — in modeled national distribution roughly 0.5% of addresses grade A and the median is C. That is a property of US last-mile infrastructure, not of your business.

Expected outage hours/year start from a national baseline of 14 hours (low end of published SMB connectivity-downtime surveys of 1–2 hours/month) and scale by redundancy, regional outage history, copper status, and environmental exposure. Where a dataset is unavailable for an address, the factor falls back to regional or national values and is labeled a regional estimate on every surface it touches.

About the outage-history index. There is no public, address-level, internet-specific outage feed (the FCC's carrier outage reports are confidential), so this factor is a modeled county connectivity-disruption index: DOE/ORNL EAGLE-I recorded power outages (county level, 15-minute cadence, trailing three years) blended with EIA-861 grid-reliability statistics (SAIFI, customer-weighted to counties), expressed as a national percentile and scaled so the median US county lands at the historical 8-event baseline (clamped to 2–24). Power loss is the dominant driver of local internet outages, but this is a power-derived proxy — never a measured internet-outage count, and it is labeled accordingly. Counties without data fall back to the ~8-event national baseline, marked as an estimate.

Datasets

FCC Broadband Data Collection wired providers per location ~biannual vintages FEMA National Risk Index hazard exposure by census tract annual US Census Geocoder / Google address normalization + geocode live, cached 30 days DOE/ORNL EAGLE-I + EIA-861 county connectivity-disruption index annual FCC copper discontinuance wire-center filings — arrives in a later release

About the provider data. Counts come from the FCC Broadband Data Collection — carrier-reported availability at advertised speeds (≥100/20 Mbps), for wired service only (copper, cable, fiber). We report it at the census-block level (the best-served serviceable location on the block), not a per-parcel guarantee, and we use only the public availability data — not the licensed CostQuest Location Fabric. FCC availability data is self-reported and, per state challenges and independent audits, tends to overstate coverage. So a count here reflects what carriers say they can offer — not delivered service or reliability — and where it errs, it generally errs toward showing more redundancy (i.e. lower risk).

Data-freshness timestamps ride along with every risk result. Addresses are used transiently for lookup and cached only as a one-way hash; a full address is stored only if you explicitly submit a form that says so.

The national ticker

The footer ticker runs at a modeled $9,041/second: ~20.5M US employer and POS-dependent small businesses (SBA Office of Advocacy profile) × a modeled mean annual downtime cost of ~$13.9k (this page’s benchmark panel at the 14-hour national baseline) ≈ $285B/year ÷ 31.5M seconds. It is a modeled rate for illustration — not a live measurement.

Estimates from public data and industry benchmarks. Not a quote. © 2026 Kajeet, Inc.