When the Power or Internet Goes Down, Where Do Your Calls Go? (2026)

When the Power or Internet Goes Down, Where Do Your Calls Go? (2026)

The cruel arithmetic of storm season is that the hours generating your highest call volume are the same hours you are least equipped to answer. Ice storms, hurricanes and grid failures produce lockouts, jammed electronic entry systems, dead fob batteries, broken keys in frozen cylinders and people locked out of homes they urgently need to get into — and if your phone system depends on your shop having power, or on one owner's cell surviving a three-day outage, all of that demand routes to whoever answered.

As of August 2026, most small trade operations have never once tested what happens to an inbound call when their premises go dark. They find out live, during the worst week of the year. This article maps every layer of a typical setup against what actually kills it, explains the one property of carrier call forwarding that most owners do not know they can rely on, and gives you a failover configuration to put in place while everything is working — which is the only time it can be done calmly.

The layer that survives, and why it matters most

Start with the single most useful technical fact in this article: carrier-level call forwarding executes inside the phone network, not on your device.

When you configure your carrier to forward calls from your business line to another number, the redirection happens at the network switch. The call never needs to reach your handset, your shop, your router, or your building. Your premises can be dark, your desk phone unplugged, your office flooded, and the forward still fires, because none of your equipment is in the path.

That is a fundamentally different reliability class from anything running on your own hardware. It is also why "forward the business line to a cloud answering point" is the most outage-resistant configuration available to a small operation — it moves the answer point off your premises entirely.

Two variants matter, and knowing the difference is what makes the setup work:

Unconditional forwarding sends every call straight to the destination, immediately. Nothing rings on your end. Commonly enabled on US mobile lines with *72 plus the destination number and cancelled with *73, though codes and availability vary by carrier — confirm yours rather than trusting a code you read online.

Conditional forwarding sends the call onward only when a condition fires: busy, no answer after N rings, or unreachable. That last condition — unreachable — is the outage case. It covers a phone that is off, out of coverage, or has no battery, and it is the one people most often leave unconfigured.

The practical setup for a trade business is conditional forwarding covering all three conditions, pointed at an answer point that is not your building and not your pocket.

What each layer of your setup actually survives

Run down your own stack against this. The failures are not intuitive.

A VoIP desk phone or softphone. Dies with your internet connection and dies with your power, whichever goes first. This is the layer owners overestimate most, because it feels like "the phone system". A well-configured VoIP account can fail over to a mobile number when the endpoint is unregistered — but that failover has to be configured in advance, in the provider's portal, and by default many are not. Check yours today.

Your business internet. Goes down with area power unless the modem, router and any ONT are all on a UPS. Note the "all" — a battery on the router alone accomplishes nothing if the modem beside it is dead. And a UPS buys minutes to hours, not days.

Your mobile phone. Survives a premises power failure entirely, which is its virtue, and fails on three other axes: battery depletion during a multi-day outage, cell-site congestion or failure during a regional event, and the simple fact that you are one person who sleeps, drives and works with both hands inside a door panel. Mobile is a good fallback layer and a poor primary one — a limitation we cover from the daily-operations angle in answering calls while driving as a mobile locksmith.

A landline or POTS line. Traditionally powered from the exchange and independent of building power, which is why it earned its reputation for outage resilience — but that resilience does not transfer to a "landline" delivered over fibre or cable, which needs powered equipment at your premises like anything else. Know which one you actually have.

Voicemail. Technically survives almost everything, since it lives in the network. Commercially it is the weakest layer in the stack, because an emergency caller in a genuine crisis does not leave a message and wait — they call the next result. During a regional outage there are always more next results. We put numbers to that failure mode in voicemail versus an AI receptionist.

A cloud answering point. Lives in a data centre with its own power and network redundancy, entirely independent of your city's grid. If your carrier forward reaches it, it answers — whether your shop has power or not.

The comparison that should drive your configuration

The rightmost column is not better because the technology is newer. It is better because it is the only column where nothing you own is in the failure path.

Configure this before the next storm

Six steps. None takes long, and all of them are impossible to do calmly during the event.

1. Find out what you have. Call your carrier and ask, in these words: does my business line support conditional call forwarding on busy, no-answer and unreachable, and what are the exact codes or portal settings? Write the answers on paper and keep them somewhere that is not a device requiring power.

2. Set the unreachable condition. This is the outage-specific one and the one most commonly left unset. Without it, a dead handset means a dead line.

3. Decide your answer point in advance. It should not be a second person's cell — during a regional outage that person has the same grid, the same congestion and the same sleep schedule as you. It should be off-premises and always awake.

4. Test it, properly. Turn your handset fully off. Call the business number from another phone. Confirm what happens. Then unplug your router and repeat. Most owners discover their VoIP account has no failover configured at exactly this moment — which is the right moment to discover it.

5. Battery-protect what is worth protecting. A UPS on modem and router, and a car charger or power bank for the phone that carries the forward. Modest cost, meaningful runway.

6. Write the storm procedure down. One index card: what to switch on, what to tell the team, who watches the incoming messages. Taped inside a cabinet door, not stored in a cloud drive you cannot reach.

Why an AI receptionist is the natural answer point

Once you accept that the answer point should be off-premises and always awake, the question becomes what it should be. A voicemail box qualifies on both counts and converts nothing. A staffed answering service qualifies but is affected by the same weather in its own region, and bills by the minute exactly when your volume spikes.

A 24/7 AI receptionist fits the failure profile precisely. It lives in the cloud, so your grid is irrelevant to it. It answers an unlimited number of concurrent callers, which matters enormously during an event that makes everyone call at once — the concurrency problem we detail in why simultaneous calls beat serial answering. It runs in English and Spanish. It takes a full intake — name, callback number, address, vehicle or lock details, urgency — and delivers each one to your phone as a structured message with a recording attached, so when your power returns you triage a queue instead of reconstructing a mystery from missed-call logs.

It also fixes the seasonal version of the same problem, which is not weather but demand: the predictable volume spikes covered in seasonal call-volume spikes and the recurring weekend gap in weekend and holiday call coverage. And because unanswered calls quietly depress the answer-rate signals that lead channels measure, keeping the line live during an outage protects more than the individual jobs — see Local Services Ads and answer rate.

A labelled hypothetical, since your figures are yours: suppose a two-day regional outage brings 40 calls you would normally handle and your line is effectively dead for both days. Even at a conservative booking rate and an ordinary lockout ticket, that is a four-figure hole — and the customers who called during it now have a competitor's number saved in their phone, which is the cost that keeps compounding after the lights come back.

Setup is deliberately trivial: forwarding, not porting. Your number stays yours, KeyBot Lite goes live in about ten minutes at $149 a month with 100 calls included and 50¢ a minute after, and you can turn the forward off whenever you want. Before configuring anything, hear it answer for your own business at the instant demo — the AI calls your phone in about thirty seconds, no signup, first demo free — or call the demo line at +1 (716) 350-6391. Plans are on the pricing page, and the always-on coverage overview is on the after-hours answering service page.

For the wider context on outage preparedness and communications resilience, the Federal Communications Commission publishes public material on network reliability during disasters.

The bottom line

Your phone system has layers, and they fail in an order most owners have never checked: a VoIP desk phone dies with the internet, the internet dies with the grid, and a mobile phone dies with its battery or with the cell site — while carrier call forwarding keeps working through all of it, because it runs inside the network rather than in your building. That single property is the whole opportunity. Configure conditional forwarding for busy, no-answer and unreachable, point it at an always-awake off-premises answer point, test it with the handset switched off and the router unplugged, and put a UPS on the equipment worth protecting. Then make the answer point something that actually converts: a 24/7 AI receptionist takes every concurrent storm call in English or Spanish, captures the full job detail, and hands you a triage queue with recordings when the lights come back. It takes about ten minutes to set up on your existing number — hear it answer for your shop first, and set it up before you need it.

Frequently asked questions

Does call forwarding still work when the power is out?

Yes — carrier-level call forwarding executes inside the phone network at the switch, not on your handset or your premises equipment, so it continues to work when your building loses power or internet. That makes it the most outage-resistant tool available to a small business, provided you configured it before the outage. The forward destination still has to be reachable, which is why pointing it at a cloud answer point rather than another local phone matters.

What happens to a VoIP phone during an internet outage?

A VoIP desk phone or softphone stops working entirely, because it needs both power and an internet connection to stay registered with its provider. Many VoIP services can automatically fail over to a mobile number when the endpoint goes unregistered, but this is a setting that must be configured in advance in the provider's portal and is often left off by default. Check and test yours while everything is working.

What is unreachable call forwarding and why does it matter?

Unreachable forwarding redirects calls when the destination phone is switched off, out of coverage, or has no battery, as distinct from the busy and no-answer conditions. It is the specific condition that covers an outage or a dead handset, and it is the one most commonly left unconfigured, which is why a dead phone so often means a dead business line. Set all three conditions — busy, no answer and unreachable — for full coverage.

How should I test my phone failover?

Turn your primary handset completely off, call your business number from another phone, and confirm exactly where the call lands and what the caller hears. Then unplug your router and modem and repeat the test to simulate an internet outage. Doing both tests takes a few minutes and is the only reliable way to discover an unconfigured failover before a storm does it for you.

Can an AI receptionist keep answering during a local outage?

Yes — an AI receptionist runs in cloud infrastructure that is independent of your building's power and internet, so once your carrier forward reaches it, it answers regardless of local conditions. It also handles many simultaneous callers at once, which matters during an event that makes everyone in your area call within the same hour, and it captures full job details in English or Spanish so you get a triage queue with recordings rather than a list of missed calls.

What does 24/7 outage-proof call coverage cost?

KeyBot Lite is $149 per month with 100 calls included and 50¢ per minute after that, no contract, and the first 5 answered calls free within a 7-day window. It runs on your existing number via call forwarding rather than porting, goes live in about ten minutes, and can be switched off at any time. The full platform with live quoting, booking and dispatch starts at $500 per month with a 14-day free trial, and all plans are published at https://www.thekeybot.com/pricing.