How to build smart home power outage alerts and backup

The easiest smart home power outage alert is usually the least trustworthy one. A Wi-Fi plug stops responding. A hub heartbeat disappears. A cloud dashboard marks the home offline. All three can look like outage detection, but they are really network-dependency detection. If the modem, router, hub, and the plug itself lost mains power together, the alert path died before it could prove what happened.

That is the useful way to think about smart home power outage alerts and backup: detection and backup are not separate projects. The alert has to survive long enough to leave the house, and the equipment that sends it has to stay powered long enough to say something more useful than “last seen five minutes ago.”

The case for taking this seriously is no longer hard to make. U.S. customers averaged 11 hours without power in 2024, nearly double the prior decade’s annual average, and hurricanes accounted for 80% of those hours, according to the U.S. Energy Information Administration [1]. JD Power reported that the average length of the longest outage rose from 8.1 hours in 2022 to 12.8 hours by mid-2025, while Utility Dive reported that 45% of U.S. utility customers had an outage in the first half of 2025 [2][3]. Oak Ridge National Laboratory estimated major-outage costs at about $121 billion in 2024 and more than $67 billion per year on average across 2018–2024 [4].

Those numbers are not the architecture. They are the permission slip. The practical question is narrower: when the power drops while you are away, which path can still tell you whether the freezer, cameras, sump pump, router, or hub are alive?

Start by separating “offline” from “power is out”

A cloud offline alert is useful after a router crash, an ISP outage, a hub lockup, or a power failure. That breadth is the problem. It tells you the service lost contact with something. It does not tell you whether the house lost utility power, whether the internet provider failed upstream, whether the hub froze, or whether the router is sitting dark because it had no battery behind it.

A defensible outage setup has two jobs. First, one signal must be able to detect or confirm the power failure without depending entirely on the dead home network. Second, the minimum smart-home stack must stay alive long enough to send alerts, log events, shut down fragile devices, and reconnect cleanly.

That table is the whole shape of the decision. A smart plug can be part of automation when the network is alive, but it should not be the only witness to the outage. A hub heartbeat can tell you the house disappeared. It cannot prove why. For a power alert you are willing to act on from another city, the evidence path matters.

Use the utility alert, but do not ask it to be a device monitor

The cheapest independent layer is the one your utility may already provide. Eversource, for example, lets customers use outage alerts and text commands such as OUT and STAT for outage reporting and status updates [5]. Other utilities use different enrollment flows, but the pattern is the same: the alert comes from outside the house, so it does not require your router, hub, or Wi-Fi to be alive.

That makes utility alerts worth enabling early. They are broad, free in many service areas, and independent of your LAN. They are also not a freezer sensor, a sump pump monitor, or proof that your rack stayed online. A utility alert can say, “the grid side has a problem.” It cannot say, “your router is still up, the hub has switched to battery mode, and the NVR has been shut down to save runtime.”

For a basic home, that may be enough to start: enroll in utility outage texts, then put the modem, router, and hub on a UPS so your normal cloud services have a chance to report their own state. The mistake is stopping at the cloud notification and calling it outage detection.

NUT is where backup power becomes a smart-home signal

For Home Assistant and similar setups, UPS telemetry is the bridge between “a battery is plugged in” and “the house knows what to do.” Network UPS Tools, usually shortened to NUT, can expose UPS state to the automation system: on battery, line power restored, battery charge, load, and shutdown conditions, depending on the UPS and integration. Home Assistant community patterns commonly use that state for on-battery notifications, graceful shutdowns, and power-shedding automations [6].

That is a better signal than a cloud heartbeat because it is specific. The UPS did not merely stop answering; it saw utility power disappear and changed state. If the router and hub are also on the UPS, the automation can still send a push notification, SMS through a configured service, local announcement, or logged event before battery runtime becomes the next limit.

The useful automations are not complicated. On battery: notify the owner, turn off nonessential PoE cameras or decorative loads, pause energy-hungry automations, and start a timer. On low battery: shut down a NAS or NVR before the UPS collapses. On restored power: notify again, wait before restarting heavy devices, and avoid bringing everything back in the wrong order.

The full Home Assistant YAML belongs in a separate build recipe, not in the middle of choosing the alert chain. If you want the automation details, use the outage automation recipe for NUT alerts, restoration notices, graceful shutdown, and load shedding. The architectural point here is simpler: UPS telemetry is only an outage alert if the path carrying that telemetry has battery-backed power.

When you need independent proof, add a battery-backed cellular monitor

A standalone cellular power monitor is the cleanest answer when you truly need an alert that does not care whether the home network survived. CabinPulse describes seconds-level outage alerts, restoration alerts, outage-duration tracking, full cellular independence, and an internal battery rated for 72 hours [7]. iSocket describes cellular power-failure alerting without Wi-Fi, with SMS, call, and email options, remote router reboot features, and prepaid plan options in the roughly $3.99-per-month class plus per-alert fees [8].

Those products solve a very specific problem: they keep watching AC power after the router has gone dark. They do not keep the router, hub, freezer, cameras, or sump pump running. Their reliability also depends on cellular coverage at the installation spot, the device battery, the subscription or prepaid plan, and the alert channel. iSocket’s own materials caution that machine-to-machine SMS reliability can be difficult in North America, which is exactly the kind of detail worth checking before treating any single box as magic [8].

Integrated alarm hubs sit in a different category. Ring says Alarm Pro can keep signaling alive for about 24 hours on its internal battery in low-power mode, adds about 8 hours per Power Pack, and provides 3GB per month of cellular backup with a subscription [9]. That can be convenient if the Ring system is already central to the house. It is still an ecosystem design choice, not a general UPS for your modem-router-hub stack.

AC power sensors that require their own hub need the same scrutiny. YoLink’s Power Fail Alarm is described as monitoring 120VAC with 24/7 heartbeat supervision, but it requires a YoLink hub [10]. That does not make it useless. It means the independence of the alert is only as good as the hub’s battery backup and outbound route.

Size the UPS around runtime at your real load

For this project, the UPS is not a vague “whole smart home” battery. It is the battery that keeps the alert path alive: modem, router, core switch if needed, hub, and possibly one or two critical network devices. The buying question is not the largest VA number on the box. It is how long that specific stack runs at its actual wattage.

Wirecutter’s measured UPS runtimes are useful because they tested the kind of small load that matters here. At a 20W modem-plus-router load, Wirecutter measured about 3 hours from the CyberPower LE1000DG, about 4.5 hours from the CyberPower CP1350AVRLCD3, and about 1.5 hours from the Amazon Basics 800VA unit [11]. Those are not promises for your house. They are a reality check against vendor runtime charts and a reminder that modest network loads can run a meaningful amount of time if the UPS is chosen for runtime rather than headline capacity.

A hub-based smart home often lands above that 20W test load once the pieces are counted. HomeTechHacker’s sizing guidance estimates a modem at about 8–20W, router at 10–25W, hub at 3–6W, switch at 5–25W, NVR at 10–30W, and NAS at 20–60W [12]. Treat those as planning ranges, not lab measurements. They are still helpful because they show how quickly a neat “router UPS” becomes a 25–100W stack after a hub, switch, and recording box are included.

This is where many smart-home outage builds get too big or too fragile. If the NVR and NAS stay on the same UPS as the router, they may drain the battery before the alert path has done its job. If the switch is omitted but the hub depends on an Ethernet run through that switch, the hub is “on battery” but isolated. The stack has to be mapped as a chain, not as a pile of devices.

If you have not measured your load yet, do that before comparing UPS models. The full measurement process, VA-versus-runtime explanation, and pull-the-plug test are covered in the smart home backup battery sizing recipe. For this alert-chain build, the short version is: plug only the required alert-path devices into the battery-backed outlets, measure their combined draw, then test the UPS by removing mains power while you are home and watching what actually stays reachable.

UPS buying filters that matter for this job

Once runtime at load is the main decision, the rest of the UPS filters become easier to keep in proportion. Wirecutter’s selection criteria include automatic voltage regulation, at least four battery-backed outlets, at least 600VA, a user-replaceable battery, transfer time of 10 milliseconds or less, and a sub-$300 price ceiling for the class they reviewed [11].

• Battery-backed outlets: count the modem, router, hub, and any switch required between them. Surge-only outlets do not help during a blackout.
• Replaceable battery: useful because UPS batteries age before the electronics necessarily do.
• AVR: useful in areas with sags and unstable voltage, not just full outages.
• USB or network telemetry: important if you want NUT, Home Assistant, or another monitor to know the UPS state.
• Runtime at your measured wattage: the deciding number.

Pick the chain that matches the consequence

A home with no sump pump, no remote freezer concern, and a tolerant owner does not need the same alert chain as a cabin, a flood-prone basement, or a house where cameras and access control matter. The right design starts with the consequence of being wrong.

For many hub-based homes, the most balanced version is utility alert plus UPS telemetry. The utility gives an outside signal. The UPS gives inside evidence. The hub can decide what to shut down or preserve. If the home is remote, high-risk, or expensive to lose sight of, the cellular monitor earns its place because it does not have to borrow the router’s last breath.

Regional risk can change the amount of runtime worth buying. A place with routine short blips may need only enough battery to ride through interruptions and reconnect cleanly. A hurricane-prone or wildfire-prone area may justify a larger tiered backup plan. Use the separate grid-risk smart home backup guide, blackout backup strategy, and hurricane prep checklist for that broader planning rather than turning the alert UPS into a whole-house battery by accident.

Do one live test before trusting the setup

The clean test is not a dashboard screenshot. It is a controlled mains pull while you are standing there. Unplug the UPS from the wall, leave the modem-router-hub stack connected to the UPS, and watch the whole chain: utility-independent alert if you have one, UPS on-battery event, outbound notification, hub state, camera reachability, low-battery behavior, and restoration notice when power returns.

Also watch the recovery order. Some devices come back before DHCP, DNS, cloud services, or the hub are ready. Others need a delayed restart. If your real problem is that devices fail to reconnect after power returns, the recovery recipes for smart devices not reconnecting after a power outage and smart homes offline after a Spectrum outage are the better place to tune dependency-order restoration.

The same test will reveal which devices actually survive on their own batteries or local protocols. For device-level expectations, keep the compatibility notes for smart home devices during an extended outage and smart home devices after a power outage separate from this alert-chain decision. The alert path has one job: stay alive long enough to tell the truth.

The final build does not need to be elaborate. Enroll in utility alerts. Put the modem, router, and hub on a UPS sized by measured runtime. Add NUT or equivalent telemetry if your automation system can use it. Add a battery-backed cellular monitor if an independent house-side signal is worth the plan cost. Once those pieces are in place, outage alerts stop being an app feature and become an architecture.

References

1. U.S. electricity customers averaged 11 hours of interruptions in 2024 — U.S. Energy Information Administration
2. Disasters Become Fact of Life for Many U.S. Electric Utility Customers — J.D. Power
3. Power outages, extreme weather, JD Power — Utility Dive
4. Analysis shows power outages cost U.S. electricity customers billions — Oak Ridge National Laboratory
5. Outage Alerts — Eversource
6. Any UPS battery backup solutions supported by Home Assistant? — Home Assistant Community
7. Power Outage Monitor — CabinPulse
8. Power Failure Alarm — iSocket
9. Using Ring Alarm Pro during power and internet outages — Ring
10. YoLink Power Fail Alarm — YoSmart
11. The Best Uninterruptible Power Supply (UPS) — The New York Times Wirecutter
12. How to Choose the Right UPS for a Smart Home — HomeTechHacker