9 mistakes I made when starting with Home Assistant

At first, home automation is exhilarating: a light that turns on “by itself”, a sensor that alerts you, a button that triggers a scene… Then reality kicks in. For us, the wake-up call was simple: when something breaks at the wrong moment (on a rushed morning, or in the evening when everyone’s tired), home automation stops being “cool”. It becomes a source of friction.

After a year and around fifty devices, I understood that the real challenge isn’t adding gadgets. It’s building a stable, understandable, and acceptable installation for everyday life: savings, security, a solid network… and zero headaches.

Here are the mistakes that cost me time (and energy)… and the choices that make all the difference for a simple, reliable smart home that stays enjoyable to live with.

❓ Why your smart home quickly becomes a nightmare

The classic trap is thinking in terms of “features”. Whereas at home, you need to think “system”.

• A system needs maintenance. The more brands, batteries, hubs and integrations you pile on, the more you increase mental load.
• A system must be robust. If Home Assistant has an issue, your home shouldn’t turn into an escape room.
• A system must be accepted. Home automation works much better if it brings comfort, savings and security, without disrupting habits.

⚠️ Mistakes I’d avoid… and what to do instead

1) Putting Wi-Fi everywhere “because it’s simple”

Mistake: multiplying Wi-Fi devices (bulbs, plugs, cameras, alarms…), then discovering the joys of network saturation, disconnections, firmware updates… and “everything depends on the internet”.

A sobering experience: during a holiday, my internet router crashed. Blackout. Immediate result: no visibility on cameras, no feedback from the alarm system, and a very unpleasant feeling of not knowing what’s happening at home. It creates needless stress… which is the exact opposite of the goal.

Instead:

• use Zigbee as the default protocol for sensors/plugs/bulbs,
• keep Wi-Fi for devices that genuinely need it,
• strengthen your network foundation (it holds everything together).

My “anti-crisis” solution, extremely practical: I got a GSM plug with a SIM card. It lets me switch the router’s socket off and on remotely, even without internet (it uses the mobile network). Honestly: it’s simple, it prevents panic, and I recommend it.

GSM Plug with SIM Card

From €30

View product

Good practice:

• everything “infrastructure” → Ethernet where possible (router, switch, access points, server, NVR),
• Wi-Fi → for mobile devices and anything that can’t be cabled.

2) Starting Zigbee without thinking about scale

With 10 devices, almost everything is “easy”. With 50, even the smallest initial choice can become a headache.

Mistake: starting with ZHA “to get going quickly”, then regretting it when the installation grows.

Instead: if you know you’re going to scale up, start directly with Zigbee2MQTT.

Feedback: repeaters are a game changer. I added 2 Aeotec Zigbee repeaters: one in the garage, and one at the far end of the house. Result: my Zigbee network became much more stable. Fewer “unreachable” sensors, less latency, less “why isn’t it responding?”.

Aeotec Zigbee Range Extender

From €30

View product

3 actions that prevent 80% of Zigbee problems:

1. Position the coordinator intelligently (not right next to the server, use a USB extension if needed).
2. Add mains-powered Zigbee devices (plugs, modules, repeaters) to strengthen the mesh.
3. Standardize your brands (see next point).

3) Spreading across too many brands — and therefore too many hubs and apps

Mistake: buying “the best product” from 8 different brands, then ending up with:

• multiple hubs,
• multiple apps,
• heterogeneous integrations,
• permanent maintenance.

My concrete case (and what I’d do differently): I started with Blink cameras (I don’t recommend them), then migrated to Tapo, then to Reolink. Along the way, I accumulated hubs: 2 Tapo hubs, 1 Tado hub, 2 SwitchBot hubs, 1 Reolink hub, etc.

It works… but it complicates everything: updates, different behaviors, accounts, dependencies, and longer troubleshooting when something breaks. Not to mention it takes up space and requires available sockets.

Instead:

• set yourself a simple rule: max 2–3 brands per category (sensors, lighting, cameras…),
• favor interoperable devices (local, standards, stable integrations),
• avoid products that lock you into a mandatory app/hub if it’s not a deliberate choice.

4) Choosing battery cameras when you want RTSP / an NVR

Mistake: a battery camera seems practical… until you need a reliable, usable, continuous stream.

Instead:

• favor wired cameras,
• check RTSP support before buying,
• if possible: a stable network (ideally wired).

Personally, I went with Reolink cameras which support this and integrate perfectly with Home Assistant.

Why battery cameras often cause problems:

• they cut the stream to save energy,
• they wake up too late,
• they don’t always provide a stable RTSP stream.

5) During renovation, not planning “smart home” sockets

This is typically the kind of mistake you regret… when the walls are already done.

Mistake: placing sockets as in a “non-automated” home, then improvising afterwards (extension cords, visible chargers, sensors impossible to power).

Instead: plan strategic sockets (sometimes at different heights) for:

• discreetly powering LED strips,
• powering presence sensors that are often mains-powered,
• avoiding visible cables.

6) Underestimating the hardware when targeting video / AI

Reality: a base like Home Assistant Green is perfect to start. But as soon as you add video, analysis, and storage, it can become too tight.

Mistake: sizing for today, then adding Frigate, cameras, and processing.

Instead:

• start light if you’re just beginning, but if video is on your roadmap: switch earlier to a mini-PC (or equivalent),
• get the server and network properly sorted (Ethernet, storage, backups).

7) Wanting to automate everything at once

Mistake: starting with “critical” automations (heating, blinds, main lights), and creating frustration at the first bug.

Instead: start with low-risk automations: if they fail, nobody complains.

Very safe examples:

• turning on an ambient light in the evening,
• “door left open” notification,
• cutting a non-essential power strip at night.

8) Not setting up “home modes” from the start

Mistake: stacking independent automations, then no longer understanding “why is this turning on” or “why isn’t this turning on”.

Instead: quickly implement 3–4 modes:

• Home
• Away
• Night
• Guest
• Holiday

9) Buying sensors with “exotic” batteries

Mistake: choosing sensors with uncommon batteries (often expensive, hard to find).

Instead:

• favor sensors with standard batteries (AA / AAA / CR2032…),
• or better: mains-powered sensors where possible (some mmWave presence sensors, mains modules, etc.).

Real experience:

I have several SwitchBot Bots, and they run on CR2 batteries. Except CR2 isn’t the battery you have in the “just in case” drawer. At the shop near me, they cost €3–4 each. When you have 2 or 3 to replace in the same week, that stings. And above all: you waste time tracking down a “non-standard” battery, instead of doing a 30-second replacement.

On the flip side: I love IKEA sensors because a good number of them run on AAA batteries. And that’s a luxury: you can find them anywhere, they’re cheaper, and you can easily stock up.

👉 Simple method to get started right (and avoid regrets)

1. Set your standards Standard batteries + default protocol (Zigbee) + max 2–3 brands per category.

2. Secure your network foundation Mesh Wi-Fi if needed, Ethernet for infrastructure, and a Plan B if your router goes down (the GSM plug is a real “safety net”).

3. Strengthen Zigbee Mains-powered plugs/modules + 1 or 2 well-placed repeaters if the house is large.

4. Deploy 3 “low-risk” automations Comfort, not critical functions.

5. Create your home modes Home/Away/Night (minimum), and base everything else on them.

6. Set up alerts Low battery, device unavailable, network loss… you want to be notified before it becomes a failure.

🩺 Common mistakes and best practices

✅ Conclusion

If I had to summarize, the real mistakes in home automation don’t come from a misplaced sensor or a misconfigured automation. They come from decisions made “on a whim”: one more brand, one more protocol, a camera “quickly picked”, undersized storage.

Over time, this creates a fragile system. Not necessarily visible at first, but which ends up being expensive: instability, maintenance, workarounds, permanent troubleshooting.

Today, I have a simple rule: I prefer a coherent smart home over one that goes in all directions. Fewer surprises, fewer dependencies, more continuity. And that’s where Home Assistant becomes what it should be: a tool that gets out of the way.

FAQ

Author: Pierre aka The Optimization Guy

Smart home enthusiast, daily optimization fanatic and tinkerer, I originally just wanted to "better manage my heating"… then I discovered Home Assistant, and I never really left.

← Previous article

How I reduced my consumption with smart plugs

📅 25 janv. 2026

No next article →

You are on the last article.