Reflections on the Smart Home Market

An essay about the current situation in the area of home automation.

This may be a quite unusual post from my side, but since I've stepped out of the home automation market 4 months ago I wanted to take the opportunity to write down my experience and insights in this special industry segment.

The area of home automation was certainly one of the first publicly known applications of the Internet of Things (or short IoT). Many companies invested quite a lot of money in harvesting these opportunities, by selling their own set of "smart devices" or "smart home solutions". There have been three kind of companies / business models that showed up particularly in that area:

  • Full solution providers (devices, sometimes with partners, plus backend - usually connected via a gateway in the user's home network)
  • Single solution providers (only one device, sometimes compatible with other full solution providers, otherwise directly connected to the Internet)
  • Infrastructure providers (concentrate on providing a backend, plus maybe a gateway to connect devices that adhere to a certain standard, e.g., Z-Wave, to the Internet)

All in all a smart home solution contains the following characteristics:

  • Devices that can talk to each other either via the Internet, via a gateway, or sometimes even in a peer-to-peer (P2P) manner
  • A rule engine to automatically trigger events when certain conditions are met
  • An interface (e.g., an app, a webpage, a digital assistant, ...) to set up the environment including the configuration of the rule engine

Note that none of these deal with actual smartness. A current smart home solution could therefore also be named connected home. Some companies try to induce some smartness, most notably Nest (owned by Google), however, the algorithms behind are mediocre at best. The Nest Learning Thermostat for instance will never be any smart without knowing more about the user(s). While the device itself may learn / repeat some patterns of its owner, it will never be able to access other (non-Nest) devices and services to be able to react smartly, e.g., to not activate when an appointment changed which will delay the user's arrival back home.

So how does the market look like in Germany? In 2010 the first solutions came out or have been planned carefully. All with high hopes mostly generated by overpaid consulting companies like BCG or McKinsey. Nevertheless, all the market projections have been in wrong and since then the high adoption is definitely lacking. One of the problems is that all smart home producers (mostly energy and telecommunication companies) fail to convey the right message to the customer. That the setup and installation is fairly easy and does - for most parts - work non-invasively. Devices like wall-mounted switches or thermostats, even smart light bulbs or smart power plugs can be installed without drilling any hole. Now that many customers are concerned that a smart home installation needs a lot of changes at their (potentially rented) home, how successful are the currently available solutions? A high installation base of any provider is between 50k and 100k customers.

What is then needed to have the originally anticipated high adoption? Certainly, a big name that can convey the message of easy and portable may help. That is why many people have been looking forward to IKEA entering the home automation market. However, at the moment the home smart series of IKEA can only be considered a start - not a complete solution. Right now it is more or less a cheap competitor to Phillips Hue, which is specialized in only lighting (smart light bulbs, switches, and plugs - granted that is nearly all you need to get started, but controlling heating and having some security aspects such as cameras, motion detectors etc. covered is also fairly important in my opinion). A killer app is still missing. Right now smart home may be more secure (or less depending on the installation), more convenient (minus changing batteries, maintenance, etc.), and even more adjustable (different colors, switches where you want them to be, unified remote controls, ...), but is definitely not a cheap endeavor. So let's make it cheaper, right?

In my opinion there are two possible ways. Either we have cheaper solutions that enter the mass market (such as cheap smartphones helped to develop the adoption of a real computer in everyone's pocket), or we have high-quality ultra premium products. Most solutions I know are directly in the middle and therefore not appealing to many people. Furthermore, we also only have the extremes on terms of usability available. We either get solutions that are only meant for experts (e.g., install the software on a Linux machine, use a Raspberry Pi, ESP, Aduino Uno etc. as basis to solder some wire) or products that will not satisfy any advanced user in terms of extensibility, hackability, or possibility of self-maintenance. In any case a viable solution must be (if wanted) much smarter than the available solutions right now. A smart home solution has to be fully secure, completely extensible, should be connectable to (most of the) available digital assistants, and provide the most convenience to users (e.g., control via voice, nearly fully available offline, availability of both - a simple and a very technical - app for all platforms including the web).

Why are the current solution providers missing most of these points? Why does the industry not come up with visionary products (any more)? Where is this heading? Well, all good questions and honestly I cannot answer any of these. I may blindly guess here and there, but I think the root cause may be a vicious circle that starting with the current solutions taking the top and saturating the existing market. Thus its hard for new solutions to come up or for startups to find a space where they can successfully operate in.

One of the obvious areas for improvement is the rule engine. Right now, open-source solutions tend to allow scripting languages or sophisticated descriptions to be supplied, while the available products only expose predefined options for creating new scenarios. The latter obviously tries to make "programming your home" easy to everyone (spoiler: it still is not - and probably never will be). I see two improvements here:

  1. Have a really sophisticated rule engine that allows an expert mode (taking input in form of a script language that was written against the API of the rule engine) as well as simple mode (mostly predefined, but everything derived from the current device model and API)
  2. Make suggestions to new rules derived either from the user's current device configuration (comparison with other / sample configurations and their scenarios) or from the user's behavior, i.e., when he uses what

Another area is the extensibility / isolation of the current products. In my opinion we need to solve three separate issues. One is the backend connectivity, essentially a relay plus data holding and prediction engine. The second one is a gateway / hub (i.e., smart home controller) to make the smart home controllable from home, as well as being able to provide communication between devices. Finally, we have the devices. This is somewhat solved (yet at least three different connectivity technologies exist, which are still quite wide-spread in the German market). Now that the devices are currently (and will potentially always be) heterogeneous regarding the supported connectivity protocol, the smart home controller has to feature (or support via hardware extensions) different network protocols. The gateway has to open a (secured) web server that exposes a standardized API. The same API should be exposed by the backend solution to which the controller is connected to. This is the client API, which can be consumed by web, desktop, or mobile applications (or other services). The connectivity to the backend may be governed by a custom API, as the connection should be configurable via an extension in the gateway. Nevertheless, from the backend's point of view the gateway should have a standardized API to fully provide the secure relay functionality. Theoretically, the client API could be sufficient, however, in practice a service API potentially needs to be added.

Why go all the way to cleanly separate these three things? Because users need to know that whatever happens to the company they trust with their smart home installation, they have the freedom of choice. If the company does not support the backend any more the users may go to another company without making any hardware change (or changing their whole solution). Likewise, if a cool new device shows up one can just buy it without having the need for another smart home controller (accompanied with a new backend connectivity, new apps, new management, ...). How would business models look like in this world? Obviously, for devices the world has always spun around innovative new use cases and improvements to justify buying a new device. Furthermore, some devices could break and need replacement. The smart home controller could be considered something like a device, plus the ability to have a market place for running (prediction or smart) services in the user's own network or when being connected to the Internet. Finally, the companies to deliver backend infrastructure would obviously do a similar thing, plus potentially charge (or need a subscription) for the remote connectivity (e.g., per month).

I still regard home automation as a really promising IoT application, however, I think it will still need some time. We are still seeing island solutions and no one of the big players is really addressing the fundamental flaws of the current products. Let's see what the future brings and when smart home installations are considered already during a standard house planning session.

Created . Last updated .

References

Sharing is caring!