AURA Devices: Growing Subscription Revenue in a Health-Tech Product
€2k → €20k MRR
Lifetime plan
Health-tech
AURA makes body composition devices: an Apple Watch accessory that measures fat, muscle, and water using bioimpedance analysis, with an iPhone app on top. When I joined, hardware was the business. The app mostly displayed measurement results, premium subscriptions generated around €2,000 per month, and every paywall change waited on developers. Over the following years I worked hands-on across research, product, pricing, and experimentation to turn the app into something people would pay for continuously.
Every user interview pointed to the same audience: fitness enthusiasts, mostly men in the US, training two or three times a week and tracking performance. To check it, I combined interviews with marketing experiments: landing pages built around different customer needs, run against real traffic. One page clearly outperformed the rest — the one about weight management, not fitness performance. Digging into behavioral data confirmed it: a large share of paying users were trying to lose weight, keep weight off, or understand how lifestyle affected their body. The most vocal users were not the paying majority.
Early conversations focused on paywall optimisation. The bigger lever turned out to be the offer itself. To make testing possible, I introduced Adapty with the team so paywalls, placements, and pricing variations became configurable without a development cycle. From there we ran experiments in 3–4 week cycles with roughly 1,000 users per variant, using total revenue — not conversion alone — as the success metric.
BIA is a fussy technology. Measurements fail if hands touch, elbows press against the body, skin is too dry or too sweaty, or pressure on the electrodes is too weak. Users don’t think in those terms. They think: it worked, it failed, the device is broken. Support data and reviews showed a pattern that mattered more than accuracy itself: people forgive imperfect accuracy, but they don’t forgive unreliable measurement.
I reframed errors from technical jargon into a human model: a clean signal has to pass through the body. We reduced the problem into two failure types users could act on — “No signal” for broken contact and “Signal noise” for distorted readings — then rebuilt the measurement instructions around posture, contact, and the moment when people get tired and lower their arms.
Then came the finding nobody wanted. Failure rates stayed high across a significant share of devices in users’ hands — regardless of instructions, posture, or UX. My experiments systematically ruled out the design hypotheses while engineers investigated construction and production. The root cause traced back to hardware, not to design and not to user behavior.
On the presentation side we tested long and short copy, photos, illustrations, image-free layouts, badges, recommended-plan labels, and button styles. Most visual “best practices” didn’t move revenue. The winning paywalls were consistently the simplest — we removed images, shortened copy, and stripped badges and decoration. The most consequential decision was the lifetime plan: the data showed monthly subscribers typically cancelled after 3–4 months, annual plans were weak, and a lifetime offer priced at almost twice the annual plan would cannibalise little. Over time, lifetime generated about one-third of premium revenue.
Results
Premium subscription revenue grew from around €2,000/month to around €10,000/month within a year, later approaching €20,000/month. The lifetime plan contributed roughly a third of premium revenue. The product evolved from a measurement display into a health and weight-management companion. Pricing and paywall decisions became a repeatable, data-driven process instead of one-off debates.
Reflection
The company ultimately did not survive: the hardware reliability problem proved fatal. The subscription revenue we built bought the company additional months of runway — design work was, quite literally, funding the payroll for a while. Two lessons stayed with me. First, the biggest monetization gains came not from redesigning the paywall but from improving the thing people were asked to pay for: the audience, the value, the offer. Second, rigorous design research is valuable even when the answer is “the problem is not in the design.” Ruling out the interface honestly, instead of endlessly polishing it, was some of the most useful work I did there.