The first time the house refused to let me in, I was holding two grocery bags and standing on my own front step. The smart lock had decided, for reasons it would not subsequently explain, that my phone's Bluetooth signature was not the one it expected. The keypad backup did not respond to my code. The mechanical key, which I had been assured was a redundant fallback, was somewhere inside the house in a drawer I no longer had access to.
I waited eleven minutes on my own porch. The lock eventually relented. I carried the groceries inside, put them away, and added a note to the spreadsheet I had begun keeping on the third day of the experiment: a record of every moment the smart home worked beautifully, every moment it failed, and every moment that fell into the strange middle category I came to think of as smart home theatre — situations where the technology functioned correctly, but solved a problem I had not actually had.
By the end of ninety days, the spreadsheet had grown to about four hundred entries. What it told me about smart homes — what works, what does not, and what is somewhere in between — was different from what I expected when I started.
The setup
For the experiment, I rented a small two-bedroom house and equipped it with the most complete consumer-grade smart home setup I could reasonably assemble. Smart locks on every door. Smart bulbs in every fixture. Smart switches behind every wall plate. Smart thermostats. Smart blinds. Smart leak sensors under every sink. Smart cameras at the entrances. A smart doorbell. A smart garage door opener. A smart oven and a smart refrigerator that the manufacturer would not stop calling "intelligent." Voice assistants in every room. A central hub to coordinate them all.
The total cost, before installation, was approximately five thousand dollars. Installation, which I did mostly myself with help from a friend who works in IT, took three full weekends.
For the next ninety days, I lived in this house, used everything as intended, and recorded what happened.
What genuinely worked
Three categories of automation produced real, measurable improvements in my daily life. I would install all three of them in any future home I lived in, even one I owned and modified at my own expense.
Lighting that responds to time and presence
The single best feature of the entire setup was, by a wide margin, the automation that turned on lights when I entered a room and turned them off when I left. This sounds trivial. It is not. The cumulative effect, over weeks, was that I stopped thinking about light switches entirely. I stopped reaching for them. I stopped fumbling at them in the dark. I stopped walking back into a room because I had left a lamp on.
Combined with time-of-day adjustments — warmer, dimmer light in the evening, brighter and cooler light in the morning — the lighting alone made the house feel calmer and more responsive than any place I had previously lived. It was the only feature of the entire system that I missed when I returned to my own home at the end of the experiment.
Leak sensors
Leak sensors are not exciting. They sit silently under sinks, behind toilets, near washing machines, and do nothing for months on end. But on day fifty-two, the sensor under the kitchen sink alerted me to a slow drip from a fitting that had loosened gradually over weeks. Without the sensor, I would have noticed nothing until water damage began to spread through the floor. With it, I tightened a connection and the problem was resolved in five minutes.
This is the kind of thing smart homes are actually good at: passive monitoring for low-probability, high-cost events. The sensors are inexpensive, the batteries last for years, and the worst case they prevent — a flood — costs more than the entire smart home setup combined.
Climate scheduling
The smart thermostat, configured to warm the house before I woke up and cool it before bed, made the temperature of the house feel correct in a way no manual thermostat ever had. I never adjusted it during the ninety days. I never thought about it. The energy bills were noticeably lower than the equivalent period at my previous home. This is the modest, slightly boring kind of benefit that does not appear in smart home advertising and that turns out to be what these systems are actually for.
The cumulative effect, over weeks, was that I stopped thinking about light switches entirely.
What was theatre
A surprising amount of the smart home experience turns out to be theatre — features that work technically but solve problems I had not had, or solve them in ways more cumbersome than the manual alternative.
Voice control
I tried to use voice control extensively for the first three weeks. By the end of week three, I had stopped using it for anything other than setting kitchen timers. The reasons accumulated gradually: voice control was slower than reaching for a switch when I was already standing next to one; voice control failed often enough to be annoying; voice control announced everything I asked it to do in a slightly cheerful voice that, after a while, began to feel like an unwelcome guest narrating my life back to me.
I understand the appeal of voice control for users with mobility limitations, for whom it is genuinely transformative. For users without those limitations, my honest assessment after ninety days is that the silent, passive automations were vastly more useful than the verbal interactive ones.
The smart refrigerator
The smart refrigerator could, theoretically, tell me what was inside it from my phone, suggest recipes based on its contents, alert me when items were running low, and play music. In practice, the cameras inside the refrigerator could not reliably identify what they were seeing, the recipe suggestions were generic, the inventory tracking required me to manually log every item I added or removed, and the music feature was worse in every respect than my actual phone.
The refrigerator was, after the smart features were turned off, a perfectly good refrigerator. With them turned on, it was a refrigerator that periodically demanded software updates and reminded me to check on its connectivity.
The smart oven
The smart oven could be preheated remotely. This is a feature I used exactly twice in ninety days, both times because I wanted to confirm that it worked, and zero times because I actually needed the oven to be hot when I had not yet decided whether to cook. The other smart features — recipe guidance, automatic temperature adjustments, app notifications when food was done — were either useless or actively counterproductive. The recipe guidance was less reliable than the recipe books I already owned. The temperature adjustments overrode my own settings in ways I did not always notice. The notifications were a less satisfying version of standing in the kitchen and listening for the timer.
What actively annoyed me
Some of the smart home setup did not just fail to add value — it actively detracted from the experience of living in the house. I will list these in roughly the order in which I came to dislike them.
Smart locks that refuse to open at the moments you most need them to open. Smart bulbs that, when the network goes down, cannot be controlled even by the wall switch. Smart blinds that decide to close themselves on bright winter mornings, plunging a sunlit room into gloom for reasons known only to the algorithm. Voice assistants that activate when you have not spoken to them, briefly broadcasting your private conversation to a server somewhere. Cameras that send you a notification every time a leaf blows past them. Software updates that arrive at unpredictable times and temporarily disable features you depend on.
None of these problems are catastrophic. But the cumulative effect of dozens of small frictions, every day, for ninety days, was a low-grade unease that I had not anticipated. The house was always almost working. The mental cost of monitoring whether it was, in fact, working — whether the lock would open, whether the lights would respond, whether the cameras were seeing things that were not there — was something I had not considered when I designed the experiment.
The house was always almost working.
What I would actually install
If I were starting over, knowing what I know now, here is what I would install in a real home I planned to live in for years.
I would install motion-activated lighting throughout. This is the single feature with the highest benefit-to-frustration ratio in the entire category. I would choose bulbs and switches that fall back gracefully to manual operation when the network is down.
I would install leak sensors under every sink, behind every toilet, and near every appliance that uses water. They are cheap, they are passive, and they prevent expensive disasters.
I would install a smart thermostat configured to follow my actual schedule and otherwise leave alone.
I would install a doorbell camera, mostly because the package theft problem in most neighborhoods has become genuinely annoying, and a camera with even moderate quality functions as effective deterrence.
And I would stop there. I would not install smart locks until they have become more reliable than mechanical ones, which they currently are not. I would not install smart blinds, which solve a problem I do not have. I would not install smart appliances, which provide trivial benefits at the cost of additional surface area for things to go wrong. I would not install voice assistants in every room, having discovered that I prefer the quiet ones.
The honest answer
The honest answer to the question of whether smart homes are worth the investment is: a few things in this category are deeply worthwhile, and most of the rest are not. The marketing presents the category as a unified experience — your house, made intelligent — and the reality is more like a collection of independent products of wildly varying quality, all sold under the same banner.
The good ones are very good. The bad ones make your home worse. The challenge, as a buyer, is that there is no easy way to tell them apart from the marketing, because the marketing for the bad ones is identical to the marketing for the good ones.
Ninety days in, I had a working theory of which was which. Three months later, after returning to a normal home, I am still using the theory. The lights still come on by themselves when I walk into rooms. The leak sensors are still under the sinks. The thermostat still adjusts itself overnight. Everything else is in a box in the basement, where it will probably stay.