I use Zigbee Lumen, Temp and PIR sensors as well as a few buttons. Zigbee is excellent for this sort of thing as the battery will often last a year or longer in a device like this.
There are advantages/disadvantages to all protocols, having a mixture of all isn’t a bad thing.
I started with z-wave, then I added zigbee devices and recently I added wifi devices. Here are my thought.
In my opinion, in 10 years, Z-Wave will no longer be there and the market will massively move to CHIP/Matter, which is compatible with zigbee and wifi.
I’m also a fan of WIFI, I have over 80 wifi devices (bulbs, sockets, switches) and counting (as well as some zwave and zigbee, mainly battery powered devices) never had any issue with them. I do have a Unifi network though so nowhere near any limits.
Have a look at Athom, they come pre-flashed with esphome or tasmota so no DIY and are WIFI and local.
I never considered the power consumption of them, would be good to see the difference in a zwave/zigbee/wifi module, has anyone ever measured them ?
Quite the opposite. Until CHIP/Matter, ESP32 and Tasmota was the closest thing to a ‘protocol’ that WiFi devices had.
THis is my favorite link on the subject at the present time:
Specifically:
(edit: another popular strategy if it works for your environment… ‘Set it all to 11.’ Note that Zigbee 11 and Wifi2.4G 11 do not overlap at all…)
Other things to watch for include USB3.0 sources. PARTICULARLY USB3.0 HDDs.
This has not been an issue for me - I have well over 50 devices on both my Zigbee and Zwave networks.
And here the setup is also quiet consistent and simple:
Lights mains powered: Mi-Light (Miboxer) - it’s own 2,4GHz network (HA LimitlessLed integration)
Switches mains powered: Sonoff WiFi
Switches and scene controllers battery powered: Hank Z-wave
Motion/temperature/illuminance sensors battery powered: Fibaro Z-Wave, Neo Coolcam Z-wave
Door/Window sensors with Open/Close state battery powered: Hank Z-wave
Door/Window sensors only motion report, but long distance thru walls, battery powered: Sonoff 433MHz
So quiet simple - mains powered = WiFi, battery powered = Z-Wave, long distance to cover = 433MHz
Only exception:
DIY Hot tub controller mains powered: Z-Uno Z-Wave
This system works like a charm, the only downtime I ever had, was each time caused by breaking changes in HA.
24x Z-Wave devices, 32x WiFi devices, 30x Mi-Light devices, 3x 433MHz devices
System is always reachable, reacts in microseconds to any trigger. Never noticed any delays.
I’m not any fanboy of any protocol. This is, how things did grow here over time. I’m just happy with the choices, I’ve made 
I think your network makes a lot of sense. You didn’t commit to a single protocol but rather the best protocol for the given task. I think this is mirrored by those that like to use Zigbee for battery operated devices and Z-Wave for hard-wired devices.
I did that with Insteon as well, at first. I kept all my wired stuff Insteon and my wireless stuff Z-Wave, but the cost and reliability of Insteon is a big factor. Having pushed through a couple hundred Insteon devices over the years, they do last for a while but when they go bad they cause the entire network to fail pretty hard and it can be tricky to nail down the culprit. That is when I committed to Z-Wave exclusively for all Insteon failed replacements or new additions.
The logic in your setup is elegant, in my opinion. While I might not choose the same brands or strategy, I appreciate the thought that went into it.
It has something on the automation side also - once finally I get the knowledge of one device, how it works, how to use it in automation’s, I get lazy to learn a second one. So if I can, I will use the same again and again.
WAF plays a role here too. If all the buttons are the same, she needs to learn only once too 
I appreciate your logic!
yup, totally agree.
I have zwave (29 devices - battery and mains), zigbee (25 devices - battery and mains) and wifi (26 devices - Sonoff, Shelly, ESP8266 DIY - all completely local - all mains powered).
I’m very happy with how the system performs and I’m not locked into (or out of) one protocol so if I have a need I can usually find it in a protocol my system supports.
I think a couple of key points on the OP’s question is that there’s no clear answer due to local dependancies. As is pointed out in this thread, radio communications performance is highly dependant on the local environment - walls, building materials, other radios, etc. The other less obvious point is that WIFI device performance is dependant on your existing WIFI infrastructure, e.g. number of devices being serviced, number of Access Points, usage patterns, etc. Zigbee and Z-wave devices create their own infrastructure so quality/performance of the existing infrastructure isn’t an issue. In both protocols success is highly dependant on how the devices are deployed, types of devices, repeaters, etc.
I’ve seen people with similar sized Z-wave networks report vastly different performance from “rock-solid, never fails” to “I’ve ripped the whole thing out in disgust”.
Plan ahead and be flexible is the best methodology.
I would.
Mentioned Shelly devices, ac or battery powered have ton of options you can configure. I don’t know any zigbee or zwave device being so much configurable. Such a flexibility is really needed.
And are not cloud based (it’s the option you can optin if you want). It has built in mqtt support so you don’t need any other integration. If you chose it provides http and coap APIs too.
Personally I have about 100 shelly devices (both ac and battery powered) and cannot say bad word about. Yes, zigbee battery powered sensors report changes faster. But slso there are zigbee devices making problems incl battery drain in days (if not used with original hub)
Wifi is also common thing we all have at home and are familiar with. Thus it’s easy to start with it. With zigbee you have may get easy start using hub (cloud you hate) or jump onto steep curve of learning about how to connect those devices locally only.
Not talking about price.
It’s fair to say that any technology has its pros and cons, It’s not nice to spread false information though.
I remember you giving me advice on Smartthings “friends dont let friends buy wifi devices”, I took it to heart and stick to zigbee and zwave. 
I will say probably the one exception i’d make to use wifi are wifi devices that can be flashed to esp or tasmota that will work 100%local with no cloud reliance. Have to be hardwired though. Id never get a battery powered wifi device.
I’m really loving this spirited debate on the pro’s and con’s of the three primary protocols!
To make it the perfect retort, it should have been followed by “… but that’s not important now.” (Those who get the reference will be ROFL. Others, please disregard.)
Back on topic, I will concede the “never WiFi” position fails with local, open-source stuff like ESP. I’ve started playing with that and no doubt will find some use cases where that’s the best solution. I have one location where I’ll want a few temperature sensors and process monitoring stuff like motor run times. It has a good WiFi signal and I can hook all the sensors and relays up to the GPIO pins on one ESP8266. That’s clearly the cheapest, most reliable and most elegant solution. It will take some DIY’ing, but no soldering required for flashing.
I have acquired all three without really realising.
WiFi
I do tend to avoid WiFi devices as there is something you can stick all WiFi devices with… They are on the same network as your personal devices and data and unless you know what you are doing they will invariably have access to the internet. So aside from the issue that a large chunk of them use basically the same controllers with very poor security my biggest concern is that if one device gets compromised (and being able to access the internet massively increases that risk) then people can get access to your network. If anyone thinks this isn’t true you only need to scan the info sec websites for regular horror stories where the attack vector was an IOT device.
WiFi also isn’t a mesh. Shouldn’t really be an issue in general but sometimes Zigbee / Z-Wave devices can build a more resilient network.
Then you have the issue that many use their own proprietary cloud services which may or not be supported in HA. A lot of devices have local versions but these are almost always not official and could stop working at any point (Look at TP Link Kasa devices).
Some devices can also stop working if your internet or WiFi goes down but this is a lot better than it used to be.
The pros of WiFi devices are that you don’t need any additional hardware, devices are widespread and cheap.
Zigbee
I inherited Zigbee courtesy of Ikea smart lights without knowing what it was. I have extended it with various temperature sensors and other things.
You need to have some sort of hardware to act as a Zigbee Hub and some way of interacting with it. I have a Conbee II USB stick and Zigbee2MQTT and Mosquitto MQTT installed alongside HA which means extra cost and complexity.
Zigbee does share frequencies with WiFi so you can get interference without a little planning.
Zigbee devices are cheap(ish) and widely available. There are plenty of devices (sensors, smoke alarms, radiator valves ,…)
Z-Wave
I ended up with Z-Wave via a Ring Alarm system.
This seems real quality but I suspect a large part of this is down to the quality of the Ring hardware rather than Z-Wave itself.
Z-Wave does not overlap with WiFi frequencies and is capable of better penetration of walls etc.
I originally intended to get a Z-Wave USB stick and connect like I do for Zigbee devices but in the end I have use a Ring integration which seems to work pretty well.
Z-Wave does have some security built into the protocol.
The biggest downside to Z-wave is cost and availability of devices. They tend to be a lot more expensive than Zigbee equivalents (but this may also be reflective of better quality)
Matter
One thing worth mentioning is there is a new kid in town. Over the next few years it looks like most Zigbee and Z-Wave devices will get slowly replaced with Matter devices (which is a new protocol that should help standardise IoT devices and means this conversation is redundant…)
This is for the future and I don’t think either Zigbee or Z-Wave are going anywhere soon. Even when Matter does come along (and some devices will be able to update to support it) I think there will be bridges between the legacy protocols and the new one
There’s a ton of strong opinions out there on these protocols. I don’t really agree with him but this video is funny (Oh the zigbee!). When he mentions Tim, I do wonder if he’s talking about me though…
And stop calling me Shirley. 
This is literally the same exception I make, Tim… The difference being it doesn’t take an act of congress to get local ESP* support on HA. Also, I only recommend it in situations where I know the person doing the install understands network engineering and the need for WiFi6 for the reasons already pointed out above, so I’ll leave that - dead equines and all.
But really, I just wanted to tell you, good luck. We’re all counting on you. 
(Please lord let’s not devolve into Python references too… If so, I call dibs on Dead Parrot…)
That video is awesome! It doesn’t convince me to go 100% wifi, but it was very entertaining to watch!
I started with zwave and decided to stick with it. It can be a bit finicky when adding new devices, but once added it’s super reliable and basically just a fire and forget thing. The things I like about zwave are the enforced compliance testing and the lower frequency. The latter is important in cases where you have challenging RF conditions in your home, like thick walls that absorb your signals. zwave has much better RF penetration than Zigbee or Wifi, that’s just physics.
What I don’t like so much about zwave are the prices. Sometimes Zigbee looks interesting due to the price difference, but I like to stick to one mesh network technology and keep it homogeneous. If you mix and match you’ll inevitably end with more maintenance and having to deal with more breaking changes and incompatibilities, etc.
Now I fully agree with all these people saying to avoid Wifi like the plague, at least for IoT. The current generation Wifi was never designed with this usage scenario in mind. If you want an even remotely serious IoT installation using Wifi, you’ll have to invest into semipro / pro network gear. With wifi the more devices you add, the less stable it becomes. With zwave or Zigbee it’s the other way round, the more devices you add the more stable the mesh becomes. Also battery life. I hate changing batteries. My zwave motion sensors are still running on the same CR123 batteries for two years now and they’re in high traffic areas. Good luck doing anything like that with Wifi.
I do use Wifi for things like my dashboard tablet though. And one security camera where I just can’t get a cable to (this makes me sad 
)
For my DIY devices I use the 433MHz band.
Edit: there’s another big advantage with both zwave/zigbee over Wifi I forgot to mention : direct association. Both protocols allow you to directly link devices together, like switches, lights, motion sensors, etc. They will still be fully controllable from HA but will also continue functioning if your HA system goes down. They allow what I like to call ‘dumb fallback’, the house falls back into a non-smart mode if the central controller fails or is turned off for some reason. Rather than becoming totally non-functional, as it would be with Wifi devices.
