I co-founded a smart home startup where we built a product that was genuinely easy to set up. Not “easy for a tech person” easy. A eight-year-old could do it. We tested it, we refined it, we got it down to minutes. And a significant number of customers bought the kit, registered an account, started paying a subscription, and never took the hardware out of the box.
The weekend they were going to set it up became next weekend, and next weekend became next month, and then it just lived in a drawer. These were paying customers. They’d already cleared every hurdle we could measure: purchase, registration, payment. But the real friction was simpler than any of that: someone had to decide to open a box and plug something in.
Even the customers who did set it up weren’t guaranteed to stick around.
Three devices in a big house
We originally built and tested the product in smaller apartments in Oslo. A gateway, a sensor, and a smart switch could serve a meaningful purpose in your living space. It felt like something. Then we tried selling to people in suburbs with bigger houses and sent them three devices. Three devices in a house with three living rooms isn’t a smart home. It’s a gadget in a corner.
When we looked at our usage data, we saw a U-curve. A lot of customers had two or three devices, basically whatever came in the box. Then a dip. Then a group who’d gone all in. Ten, fifteen, twenty devices. The middle was almost empty.
The threshold was around six devices. That’s where the system started behaving like a system. Multi-room control, proper monitoring, automations that actually made a noticeable difference to daily life. Below six, you had a product. Above it, you had a smart home.
We used to say that to understand the value of a smart home, you have to have a smart home. The obvious question is why we didn’t just sell bigger starter kits. The answer was price. A six-device package costs roughly twice what a three-device one does, and you’re asking someone to spend more on a product whose value they can’t yet see. That’s the paradox, and we, honestly that entire industry, never really solved it.
The false summit
Startups tend to pour their energy into the first aha moment. Get users to value faster. Reduce time-to-activation. Smooth out onboarding. It’s important work, and it’s tempting work, because it’s easy to measure, easy to test, and easy to show progress on. But at some point you’re optimising something that’s already good enough while the real problem is somewhere else entirely.
When you build something and pour yourself into making it good, there’s a natural tendency to treat activation as validation. Someone is using the thing you built. But the question isn’t whether they used it. It’s whether they used it enough to see what it actually does.
One deal in a CRM is a sample. Fifty deals with your whole team logging activities is when the tool disappears into the workflow. Most products have a version of our six-device threshold, a point where individual usage tips over into something genuinely sticky. Most companies aren’t tracking where that point is.
The data you’re not looking at
What made this harder than it needed to be was that our instrumentation wasn’t built to find these patterns. The data existed, sitting in NoSQL databases, technically accessible but practically invisible. We tracked what we needed to ship the product and keep it running. We didn’t track what we needed to understand adoption.
This is common, especially at startups. Early on, the CTO sets up database structures from a technical perspective, because that’s what matters at the time. Nobody’s thinking about what a PM or growth person will need to ask of that data a year later. By the time you start looking for deeper adoption patterns, the data is spread across systems that were never meant to talk to each other.
The six-device threshold was in our data the whole time. We just took too long to find it.
What we’d do differently
We were a white-label platform. We didn’t control the pricing, the packaging, or the sales conversation. We couldn’t have just sold bigger starter kits or offered installation services. The lever we actually had was the partnership: structuring the relationship so that data flowed both ways.
If we were starting again, we would have built that into the contracts from the beginning. Sales data, activation data, expansion data. Not as a reporting requirement, but as a shared investment in understanding the customer journey together. In a B2B2C relationship, the data isn’t yours or theirs. It’s about joint success. Without it, we were spotting patterns too late and had no way to act on them even when we did.
Start before it feels necessary
When you have twelve customers, none of this feels relevant. You know their names, you know what they’re doing with your product, you can call them up and ask. Instrumentation feels like overkill when you can just look.
But that’s exactly when the structure needs to go in. Not because the data matters yet, but because the habits and the infrastructure you build at twelve customers are what you’ll be stuck with at twelve thousand. By then, the questions you need to answer are harder, the patterns are subtler, and the cost of not having clean data is measured in months of delayed insight. The second aha moment was in our data the whole time. We just didn’t build the system to see it until it was too late to act on it.