Oura has been making waves (again!). The company is clearly doing well overall: They’ve sold over 5.5m rings, claimed roughly $2 billion in projected 2026 revenue, hit an $11 billion valuation, and confidentially filed for IPO.
Pretty mediocre timing to be hit with a class action lawsuit alleging false advertising. According to the lawsuit, the ring’s sleep measurements are “AI-generated guesses” with “a coin flip’s chance” of being right. The advertised “95% sleep staging accuracy compared to a clinical sleep lab” is not backed by evidence.
Should we believe them? Is Oura wildly inaccurate and mostly built on marketing?
I sense checked the studies both sides quoted. And not only that, I also dug into the LinkedIn debate that ensued between the Oura leadership and external scientists. Yep, I don’t mind putting my life on the line for y’all. In the trenches each week.
Let’s discuss which claims on both sides are valid, and how on earth both fit together.
Shoutout btw to Olimpia Lopez Faba for inspiring this post! She has a science background and writes about wearables regularly. She sent me some smart takes on the topic, which I’ll quote below.
But first, some sleep basics
To understand the debate, we have to align on some things about sleep.
The gold standard of measuring sleep is polysomnography (PSG): a lab-based overnight test where electrodes on your head and face record brain waves, eye movements, and muscle tension. A trained specialist then looks at the recording in 30-second snips and labels each one into the sleep stages. The stages are roughly speaking being awake, in light sleep, deep sleep, or REM.
Accuracy in this field usually measured by asking: In what share of the 30-second clips does the device’s label match the human’s label?
Core of the definition are the brain and facial signals. Deep sleep means slow brain waves, REM means rapid eye movements. A ring on your finger obviously can’t record that. That’s not a show stopper per se, but it means a ring like Oura needs to rely on “side effects” (such as heart rate, heart rate variability, breathing, temperature, movement) to define sleep stages. It guesses what your brain is probably doing, very very indirectly.
Two things to keep in mind. First, there are two very different measurement tasks: Deciding whether you’re asleep at all (easy) vs. deciding which of the four stages you’re in (harder). Second, a device can properly measure totals across people and sleep cycles, but still be wrong about individual times. Errors in both directions cancel each other out - so we’re working a lot with averages here.
Where the lawsuit is right
Let’s start with the criticism. These are the lawsuit claims I’d underwrite:
Oura markets 95% accuracy, which is misleading. In their marketing material, they suggest “95% accuracy” for overall sleep measurement. That is cherry picked: 95% is their accuracy for deciding between asleep and awake. The more interesting number, sleep staging accuracy, is only 79%. One scientist on LinkedIn phrased it perfectly: “Nobody buys a ring to answer ‘am I awake or am I sleeping?’”
The one truly independent study on Oura produced bad results. All of the studies Oura likes to quote are paid by Oura and/or have Oura affiliates involved. One independent study exists: Researchers at Charité Berlin tested the ring on 45 sleep-lab patients (sleep apnea, insomnia, narcolepsy). In the study, four-stage accuracy reached only 53% and the ring caught only 46% of the moments patients were actually awake. It overstated REM by 32 minutes on average. That does not look good in the absence of other independent evidence - although there’s limits to this study, which we’ll discuss later
The population in Oura’s studies is… selected. Favorable studies on Oura like to exclude people with sleep disorders, high BMI, and even caffeine use. But a natural customer of a sleep tracker is someone worried about their sleep and might have a condition - are we sure it holds up in these groups? Olimpia had a good point here:
“In clinical science, we report performance by cohort because physiology determines reliability. Consumer wearables should adopt the same standard, publishing accuracy matrices by physiological state rather than universal averages”
Oura’s (overall solid) defence
Oura managed the public response against these allegations quite well, in my opinion. They responded along the scientific rules, meaning they quoted evidence supporting their points and tried to methodically debunk their opponents. All fair game.
Specifically, their strongest points:
The lawsuit’s anchor study is flawed, which Oura dissected in a formal letter to the journal: It uses data from 2022 on an older ring and algorithm, and comparisons were done using 5-minute windows instead of the 30-second chunks the algorithm was built on. That mechanically reduces accuracy. They also crowded three rings onto each patient’s fingers
The gold standard itself is vague. When two trained experts in PSG score the same data, they only agree about 83% of the time. For deep sleep specifically, expert agreement averages only 67%! Against that ceiling, 79% on healthy people is close to the best physically achievable. In other words, it matters whom you’re competing against…
Oura wins every head-to-head wearable comparison, including the independent ones. In one of their own studies Oura beat Apple Watch and Fitbit, and even in the Charité study cited by the plaintiffs it was the best of the three rings tested
The lawsuit argument “No brain electrodes, therefore guessing” is nonsense. This was the lawsuit’s weakest claim. A solid body of research shows that heart rhythm, breathing, and temperature do actually change in characteristic ways across sleep stages. They’re linked. Plus, the gold standard (PSG) itself is not an accurate measurement either! It’s just a combination of proxy markers like EEG. To quote Oura’s head of science: “Sleep is not only EEG, and EEG is not a direct measure of sleep either. It is one window into it”
The plaintiff’s narrative has another basic intellectual flaw: Calling the ~53% a “coin flip’s chance of being correct” is statistically absurd. The algorithm needs to pick between 4 sleep stages, so guessing between them would be 25% accuracy... the lawsuit was clearly not written by scientists.
How do both sides fit together?
As we’ve seen, both sides have valid numbers, and none of them are fabricated or fraudulent. The gap comes from Oura making some very deliberate choices to appear in the best possible light:
On the communication level: They generalised, placing the 95% stat (“asleep or awake”) in a way that suggested it represents overall accuracy. They’re also opportunistic with picking benchmarks: Oura blames the vague gold standard to excuse its 79%. Fair enough, but then why claim “95% vs. a clinical sleep lab” in theirs ads?! It should cut both ways…
On the scientific method level: They’ve used the common levers most companies would go for. They’ve reported greatly accurate averages although an inidividual customer might still get nights with errors of 100+ minutes. And they’ve selected user populations in their favour
I must say that overall Oura’s science looks solid. They’re in an incredibly difficult research field with a bad gold standard. The ring’s sleep measurements are not super accurate, that’s just a fact. But for the overall state of the field, they’re doing well.
As someone with a medical background, I’m just disappointed with their marketing. They obviously put the best numbers forward, likely knowing that people won’t differentiate. As excellent researchers, shouldn’t they pay tribute to complexity?
To quote Olimpia’s more scientific framing:
“When a wearable frames its outputs as near clinical accuracy, it creates a psychological asymmetry: users begin treating the device as a superior interpreter of their internal state. High‑certainty metrics reduce autonomous interoceptive interpretation and increase cognitive outsourcing. In practice, this means users become more anxious, more dependent and less capable of contextualising their own signals.”
In other words, your smart ring might be gaslighting you.
Bottom line and a look forward
Companies like Oura should just be open with the fact that they’re mostly estimating/guessing, and that’s fine. It’s how medical science often works. At the same time, consumers should be educated around the uncertainty in medicine instead of being given a false sense of accuracy.
This whole story also teaches a side lesson for investors: Many of the newer tracking devices (rings, wristbands, micro-needle patches) operate on inherently inaccurate technology. Most of them are in the business of estimating, especially the non-invasive ones, and are likely tempted to oversel their accuracy. Pragmatically as an investor, I’d check a) how strong is the standard of care and b) whether they focus on consumers or a medical market. The weaker the standard of care and the more it’s a consumer play, the better your odds of success. Oura is the living proof.
So what’s next for Oura? I’m certain they’ll recover from the lawsuit. They will have a strong IPO. Also, the LinkedIn debate revealed an interesting hint at Oura’s future. The CEO left a comment that sounded like Oura is planning to become a medical device!
That would mean a huge increase in trust for the brand. It would finally give them full medical legitimacy. It would also endanger a number of startups in the field, especially those whose whole pitch is being the clinical-grade wearable, and love to look down on Oura.
Exciting times ahead!
Speak soon,
Lucas
P.S. does anyone have access to Oura secondaries? asking for a friend
Honoured to have contributed to the analysis of a case that will set an important precedent in the healthtech ecosystem.
These moments expose the tension between physiological truth, scientific rigour and commercial narrative and they force the sector to mature.
At The Centenarian's Path, I work precisely so cases like this become usable frameworks for companies: clear signals of what stands, what fails and what must evolve.