How (Not) to Introduce a Subscription Fee Where There Was None

Some products where there is both a hardware and software component will charge for the hardware, and then continue to enhance the software for “free” (i.e. include any running costs in the original hardware purchase price). But what happens when it transpires that the “free” services are costing the company too much to operate?

Netatmo (not Tado!) smart thermostat.

Tado: Smart Heating, Less Smart Clawback

I was recently in the market for a smart thermostat, and researched a number of vendors. I was about to go with Tado, who have a nice-looking set of hardware products, when I noticed various forum posts about their new charging model. Previously they had sold (fairly expensive: £60/$80 per radiator valve) products, and then provided the control software for free. Recently, they announced that they would charge for some software features, at roughly €3/$3 per month.

What had really irked the user base was that (a) one of the most-coveted features (geofencing to turn automatically heating on when the user is heading home) had become a subscription feature, and (b) future software features would now be available only on the new version of their application, and would require a subscription.

Now, keep in mind that this change was made to customers who had already made significant purchases with Tado. If this had been a free piece of software that had suddenly become chargeable then it would have been different, but customers had been sold hardware with free software, only to then be told to subscribe.

Why the Change?

(Disclaimer: I don’t have any inside information on Tado, hence I can only guess as to the true reasoning; the below is my personal opinion and conjecture.)

If you’re in the Software as a Service (SaaS) business you’ll be familiar with that it’s great because you can charge a subscription which nets a predictable, regular income. However, you’ll also be familiar with the issue that this income arrives after you’ve made the outlay in infrastructure to service the additional business: even if you take a month’s payment up-front, it’s highly unlikely to cover your investment: that will take several payments. Of course, over time the subscription will more than pay for the investment (and thus profit will result), but you’ll be constantly “behind”. Of course if you don’t charge a subscription at all, you’re relying on your up-front payments to cover your costs forever…

I’m not certain what the margin on Tado’s hardware products is, but 10% probably isn’t a crazy number. If you invest $800 on a system, we’ll assume that they make $80 of profit. But that “profit” then has to pay for any running costs, e.g. web hosting, and application development. Moreover, let’s assume that such hardware will have to have a lifetime of at least 5 years to justify someone installing it (I mean, how often are you likely to replace your smart radiator valves?!). That works out at $16 of margin on the hardware for each year of usage of the product, per customer.

A “normal” sized (m5 on AWS) instance is priced at roughly $0.10 per hour. Now of course one can get discounts for reserved instances etc., but we’ll ignore those for the moment. A year’s-worth of a single AWS VM is therefore over $800. Clearly we’ll need at least two of these (and let’s hope there are few more in case an AWS region goes down), and we might need to pay for some egress network traffic, but let’s be generous and put the yearly bill at $2,000.

That’s the equivalent of 125 customers (all at $16 a year). With zero left over for paying for application development, or (shock, horror) actual profit for shareholders.

Meanwhile, if Tado charge every customer $3/month, that’s $36 a year, which gives us $20 a year per customer to spend on things other than AWS, or take as profit.

Scale is Your Friend

Of course, at some point the costs of running the service become small for each additional user: if you have 10,000 users (and we assume we can service 500 per VM, plus we have 25% redundancy), we’ll need somewhere in the region of 25 VMs, at a yearly cost of $20,000, whilst our $16/user/year provides us with $160,000, leaving $140k to spend elsewhere.

(By the way, $140k/year for 10,000 users sounds very low to me. A consumer software subscription like Spotify might be $10/month, with a greater than 20% margin, meaning a profit of $25/user/year, or $250k/year.)

You can see why Tado’s adopted this approach. Until they can grow to having a significant user base, the costs of servicing it are likely to chew up any profits very quickly, and prevent them from investing in significant developments in their software. By the way, the software is almost certainly where they will be able to differentiate from their competitors (radiator valves are, well, radiator valves, after all). And differentiation is how to gain customers.

Solution: How Does a Company Introduce a Subscription?

Let’s assume that we’re Tado and we’ve now realised that we need to charge a subscription in order to be able to grow the business sustainably (i.e. not require further money from investors, or take on debt). What’s the best strategy to make this happen?

Clearly you don’t want to upset your existing customers: they like your product, and given that they’re early adopters, are probably your biggest proponents. And we all know that a personal recommendation is by far the best way to capture new customers. But we need them to start paying us something: they’re the ones that cost us the bulk of the running costs, after all.

Firstly, if someone has paid for something, it’s unwise to try to charge them again, particularly not for the killer feature in your product. In my view this was Tado’s biggest mistake.


  • Invest in creating several new differentiating features, and introduce these into a new premium tier of the application.
  • Make very clear that the free tier will always be available.
  • Ideally, add a (less significant) new feature or two to the existing free tier, to make it clear that you intend to continue developing it.
  • Finally, offer existing customers the ability to purchase the subscription at a lower rate than new customers (cover your costs, rather than profit).
  • If possible, offer access to the product’s API (to anyone, including new customers) for the price it costs to provide that API service. This allows early adopters who are happy using an API to attempt to recreate your premium features. (Of course, this only works if the API isn’t already available for free!)
  • Over-communicate about the change: make sure that new customers understand the tiers of product available to them, and that existing customers are told (and see evidence of) continued development of the free tier, as well as the availability of new value-add services.
  • Be prepared for current customers to be upset and confused. Tado blogged about the change, but it was hard to find the page. They then listened to feedback and revised the blog, offering existing customers promotional pricing on the new subscription (but still charging for the geofencing feature which had previously been included in the hardware purchase price).

To Tado, or Not to Tado

Of course, some of you may be wondering what smart heating system I went for. The suspense can now be lifted: I went for a system from Netatmo. The costs worked out to be very similar for the hardware, but Netatmo make clear that they have no intention of charging a software subscription.

If Tado had made clear on their web site that it was a subscription product, I might have gone with them, despite the additional cost: they appear to have good ideas, and well-designed hardware. But the fact that there is a subscription is absent from many pages, even their web shop for the hardware does not mention it. As much as I would not have been happy with the introduction of a subscription as an existing customer, that would not have affected my opinion as a new customer. Moral of the story: make it clear what your pricing is, if you require a subscription.

Two levels down in the Tado site, if you look closely, you’ll find you have to pay a subscription (as indicated by my added red arrow). Not exactly super-clear, and certainly not on the front page! (Note: actual page has various videos on it that haven’t survived the screen capture process — apologies!)

Now, Netatmo may yet make the same change. I’m hopeful they won’t, in part because they appear to have a good number of existing customers (see their map of customer weather stations, for example), and a range of other products. That scale hopefully means that they’re able to service the costs of their customers whilst making a decent profit. Time will tell!