Your first five smart home devices, in order
The order matters more than the brands. Buy a spine, then one room, then a physical control, then insurance.
Why order beats brand#
Most first smart homes are bought as a pile: a four-pack of Wi-Fi plugs, some color bulbs, a video doorbell. Everything works individually and nothing works together, because nobody decided where automations would run or how devices would be named. Two months later the plugs are unplugged.
The order below fixes that by front-loading the decisions that are expensive to reverse. Devices are cheap and replaceable. The controller, the radio you commit to, and the naming scheme are not. Read which radio for which job once before you spend anything, and run the platform picker if you have no existing loyalty. Both take ten minutes.
Buy the controller first, and make it a border router
This is the only purchase that constrains everything after it. You want one box that is both a Matter controller (it owns the devices and runs automations) and a Thread border router (it connects battery Matter devices to your network). Most households already own a candidate: a HomePod mini, an Echo with Thread, a Nest Hub, a SmartThings Station, or a Home Assistant Green with a radio dongle. Check the specific model, because near-identical products differ: the 128 GB Wi-Fi plus Ethernet Apple TV 4K 3rd gen is a border router and the 64 GB Wi-Fi only model has no Thread radio at all. The border router finder filters every shipping US model by ecosystem.
Buy one smart plug and put it on a lamp you already use
A single plug, around $15 to $25, is the cheapest possible test of the entire chain: pairing, naming, app control, voice control, an automation, and what happens when the internet drops. Put it on a lamp you switch on every evening, not on something decorative. Within a week you will know whether commands feel instant or laggy, whether voice hears the name you chose, and whether your household tolerates it. Buy Matter certified rather than a vendor-app-only plug, so it can move to a second platform later. Ignore the energy monitoring numbers for now; accuracy at low loads varies widely between models.
Do one room of lighting properly
Pick the room you spend the most time in and light it completely, rather than putting one smart bulb in five rooms. Completely means every fixture people use in that room, so nobody ever has a reason to say "the smart lights do not work in here". The bulb-versus-switch decision is the whole game: if the room's light is controlled by a wall switch that people flip, a smart switch with dumb bulbs is more reliable, because a smart bulb with the power cut is a brick. If the room is lit by lamps and floor fixtures, bulbs win. In older US homes, check the switch box for a neutral wire before you commit to a wired switch, because many boxes wired before the 1980s have only the switched hot and ground. Smart switches, dimmers and the neutral wire and smart lighting that stays compatible cover both paths.
Add a physical control that does not depend on the app
The fourth purchase is the one people skip, and it is the reason their fifth purchase never happens. A battery-powered wireless button or dimmer, stuck on the wall by the door, gives everyone in the house a way to control the lights that does not involve unlocking a phone or being heard by a speaker. It also means a hub outage is an inconvenience rather than a crisis, particularly for anyone who did not ask for a smart home. Where the room uses smart bulbs, mount the wireless control over the existing wall switch so the switch cannot be flipped and cannot cut power.
Buy a leak sensor and put it where water actually appears
A $20 battery leak sensor under a water heater, behind a dishwasher, or beside a washing machine is the only device on this list that can pay for itself in a single event. It is also the easiest automation you will ever write, because it has no design questions: water, alarm, notify, done. No presence logic, no timeouts, nothing to annoy anyone. If you buy nothing else from this page, buy this. Sensors: motion, contact, temperature, leak covers battery life and reporting behavior.
What each purchase is really buying#
| Step | The device | What you are actually buying |
|---|---|---|
| 1 | Controller and border router | Where automations run, and whether they survive an outage |
| 2 | One smart plug | A cheap end-to-end test and your first naming decision |
| 3 | One room of lighting | Daily value, and proof the household will accept it |
| 4 | A wireless button or wired dimmer | A fallback, and consent from everyone else in the house |
| 5 | A leak sensor | Insurance with no automation design required |
Total is roughly $150 to $350 depending on how the room is lit and whether you already own step one.
What not to buy first#
- A smart lock. Locks fail in the most socially expensive way possible: someone stands outside in the rain. They also involve credentials, guest access, battery management and often the door hardware itself. Buy a lock once you trust your network, not before. Smart locks and the Aliro standard explains what Matter does and does not expose.
- A camera or video doorbell. Cameras are the category where Matter has helped least. Matter 1.5 defined camera device types in November 2025, but as of September 2026 SmartThings is the only major ecosystem where a Matter camera actually works, and the specification does not cover storage of footage at all, so a subscription can still be required. A first camera is a purchase into a vendor's cloud, not into your smart home.
- A starter bundle with its own hub. Bundles are priced to lock you to one vendor's radio and one vendor's app. You end up with a second hub before you have a first system.
- A thermostat. Not because thermostats are bad, but because they are a wiring job, they interact with HVAC equipment you may not understand yet, and the failure mode is a cold house. Also check for a C wire before you order anything.
- Twelve Wi-Fi devices. Every Wi-Fi smart device is a client on your 2.4 GHz radio competing for airtime with your laptops. There is no credible published limit for a consumer router, but the constraints are real: DHCP pool, per-radio association table, and airtime. Past roughly 50 Wi-Fi IoT devices the answer is different hardware or a different radio. Wi-Fi smart devices: the hidden costs explains the mechanism.
Three decisions to make before step one#
Where automations run. If they run in a vendor cloud, an outage takes them out. Google's global outage on 18 August 2026 lasted about twelve hours and left Nest hubs unresponsive. If they run on a local hub, they keep firing. This is the single biggest difference between platforms, and local vs cloud control lays out the five places a command can be processed.
Whether anyone else in the house has a veto. They do. Design for the person least interested in this, which is what step four is for.
Your naming scheme. Decide it before device two, not after device twenty. Names should be things you would say out loud in a sentence. Naming and room structure that scales is a fifteen minute read that saves an entire evening later.
After the five#
Stop and live with it for a month. The correct sixth purchase is almost never predictable in advance, because it comes from an annoyance you have not had yet. When you do buy again, run the item past the pre-purchase checklist first.
If you are renting, the whole sequence still works, but steps three and four change: see smart home for renters for the versions that leave no holes. When you do start writing automations, automations people do not turn off covers the design rules that decide whether they survive contact with your household.
Do I need a hub for my first smart home?
Not always. If you buy only Wi-Fi Matter devices and control them from phones, no hub is required for local control on your own network. You need one the moment you want battery Thread devices, automations that run while your phone is elsewhere, or remote access. Most people reach that point within a few months, which is why buying the controller first saves a re-do. Do you need a smart home hub? has the four honest answers.
Should my first devices be Matter certified?
Where a certified option exists at a sane price, yes, because Matter certification is what lets a device be shared to a second ecosystem later without rebuying. Be careful with labels: only certified devices carry the Matter logo, and the "works with" badges on boxes are ecosystem programs that test the platform experience rather than Matter conformance. Works with Matter vs Matter certified decodes the box.
Is it cheaper to buy a big starter bundle?
Per device, usually. In total, usually not, because bundles push a proprietary hub and a fixed device mix that rarely matches your rooms. You also lose the most valuable part of the staged approach, which is finding out what annoys your household before you have bought twelve of something.
Smart bulbs or smart switches for the first room?
Switches if a wall switch controls the light and people use it, because a smart bulb loses power when the switch is off and then appears offline in every app. Bulbs if the room is lit by lamps, or if you rent and cannot touch wiring. Color capability is the tradeoff: smart switches give you reliable dimming of dumb bulbs, not color.
How many devices before things start breaking?
Symptoms usually start with radio design rather than device count. A Thread network allows 32 active routers, and battery devices never relay, so a house of battery sensors with one mains-powered device has no mesh at all. On Wi-Fi, congestion shows up long before any documented device limit. Adding a couple of mains-powered relay devices is a better fix than adding a second hub.
What is the cheapest useful smart home?
A leak sensor and a smart plug, roughly $40 together, with control from a phone you already own. That is a real smart home: it prevents one expensive failure and automates one daily action. Everything past that is a preference, not a requirement.
Primary sources
Specification and vendor documentation we checked while writing this page. Where a claim depends on firmware behaviour rather than a published spec, the page says so inline.