What You Didn’t Know About Your Own Body

We tracked over 3,000 users through January to measure whether the resolution effect is real. It is. Steps jumped 13.3 percent on January 1st, a statistically significant spike. But the median duration was just two days. By day three, most people were back to baseline. 80.6 percent managed only one to two days. Fewer than 3 percent made it past five.

The surprise was who tried hardest. The least-active quartile showed gains of 25 percent, while already-active users barely moved. The people with the most to gain vanish fastest.

We mapped 8,298 sleep records across 15 countries and found that average sleep ranged from 7.3 hours in Ireland to 5.8 hours in Indonesia, a gap of more than 90 minutes within the same month. Northern hemisphere countries in autumn logged longer nights, likely driven by cooler weather and shorter days, while equatorial regions like Indonesia recorded the shortest sleeps. New Zealand led the southern hemisphere at 7.4 hours despite heading into spring.

The variation is not just about duration. Night-to-night consistency differed dramatically. The UK and US had the most erratic sleep, with standard deviations above 3 hours, while Austria and Ireland were the steadiest sleepers. And weekends reshuffled everything: US users actually slept less on Friday and Saturday nights, while Italian users trended half an hour more. Same biology, completely different sleep cultures.

Most people assume alcohol destroys their sleep architecture, knocking out REM or deep sleep entirely. Our research found something more nuanced. Sleep stage composition didn't shift dramatically, though lab studies with controlled doses tend to show clearer effects than the messy variability of real-world drinking. What did show up clearly is fragmentation. Awake time increased by about 5.5 minutes on average, with a sharp rise in wakeup events later in the night. Breathing rate also rose by 0.17 breaths per minute. You get a night that looks normal on paper but feels terrible in the morning.

The practical takeaway is counterintuitive. The biggest effect of alcohol on sleep in real-world data isn't the stage-level disruption you'd expect. It's fragmentation. You spend more time awake in bed, you wake up more often, and your body lingers in light transitional states where it isn't really recovering. The total hours look fine. The quality underneath is quietly degraded.

Everyone assumes melatonin is a sleep aid. We looked at over 200,000 nights of sleep data and found zero measurable effect on how long people sleep, how fast they fall asleep, or when they drift off. Not a small effect. Zero. What melatonin actually does is something nobody talks about: it lowers your resting heart rate and raises your heart rate variability, peaking around days 3 to 5 of consecutive use. After that, both metrics drift back toward baseline. By about day seven the effect is mostly gone, and there is no withdrawal or rebound when you stop. Your body just returns to normal within a couple of days.

The practical takeaway is simple. Melatonin is not a sleep pill. It is a short-lived shift in your autonomic nervous system, a temporary nudge toward recovery that wears off quickly. If you are taking it expecting to sleep longer or fall asleep faster, the data says you are solving the wrong problem. But if you think of it as a brief recovery window, something that gives your cardiovascular system a few days of improved tone, then the data actually supports that. Just do not expect it to last.

We analyzed 59,000 daily records from 256 users and found something most people would not expect: sauna days produce a lower nighttime heart rate than exercise alone. On days when someone used a sauna, their minimum resting heart rate dropped by about 3 bpm, roughly 5 percent, and that effect held even after controlling for the fact that sauna users also tend to be more active. The drop is not because they exercised harder. It is a distinct physiological recovery signal, likely driven by increased parasympathetic tone during the post-sauna cooling phase.

The effect is consistent through the night. In the first hour after sleep onset, heart rate on sauna nights rapidly decreases creating a gap between sauna and non-sauna nights of 3 to 5 bpm. For women, the story gets more nuanced. The heart rate benefit only appears during the luteal phase of the menstrual cycle, with no meaningful effect during the follicular phase. This suggests the recovery mechanism interacts with hormonal state in ways we are only beginning to understand.

We studied 538 users tracking daily steps, heart rate, and HRV around their GLP-1 doses. Activity drops on medication days and stays suppressed for up to seven days. Resting heart rate rises slightly but persistently, peaking two to seven days post-dose. HRV drops and stays down for weeks. These are within-person comparisons, each user measured against their own baseline, so this is a population-level trend, not noise.

What surprised us most was what exercise could and could not fix. Fitter individuals showed smaller heart rate increases, but HRV did not follow the same pattern. It dropped and stayed suppressed regardless of fitness level. Heart rate and HRV are telling two different stories.

We analyzed over 70,000 activities from 305 GLP-1 users. Before starting GLP-1, the pattern is what you would expect: more exercise predicts lower nighttime heart rate and higher HRV. After one month on GLP-1, that relationship inverts. Additional exercise is now associated with lower HRV, not higher. The exercise-recovery coupling has fundamentally flipped.

Baseline fitness partially offsets the heart rate increase, fitter people show a smaller rise, but HRV stays suppressed regardless. There also appears to be a sweet spot: small increases in training load, around 30 minutes above baseline, show minimal impact. Push beyond that, and heart rate climbs while HRV stays lower.

This is one of the most counterintuitive findings in endurance training. Runners who spend more of their training time at an easy, conversational pace finish marathons significantly faster than those who push hard in every session.

The runners who hammer race pace in training actually get slower. The more time they spend in Zone 3, the worse their finish time. Meanwhile the ones who take it easy for most of their training and only go hard occasionally are the ones crossing the line first.

Norway took 41 medals at the 2026 Winter Olympics, 0.73 per 100,000 people. The United States, with 60 times the population, managed 0.01. The obvious explanation is elite training programs. But when we looked at over 10,000 European users' activity data, something interesting showed up. The average Norwegian logs 512 minutes of activity per month, 39 percent more than the average user. They walk and hike 64 percent more. They cross-country ski 20 times more.

Even the least active Norwegians spend more time in low-intensity zones than low-activity users elsewhere. The whole population moves differently. We hypothesize that this leads to the largest pool of talent at the base of the talent pipeline pyramid and demonstrates that a strong culture that emphasizes being outdoors and moving positively impacts elite performance.

We built a model that explains about 79% of the variance in marathon finish times across 101 runners. It uses five inputs: your baseline pace, total training volume, how you split your intensity zones, and how often you run per week.

The single biggest predictor is your baseline pace. It accounts for more than half the model's power on its own. After that, the split between easy and hard running matters more than most people think. Spending 75 to 85% of your volume in zone 1 produces outsized improvements. Going hard more than 20% of the time actually makes you slower.

This calculator is a rough prototype, not gospel. It's off by about 18 minutes on average. But it's a useful way to play with the tradeoffs and see how different training choices might shift your finish time. It refers to the 6 months of training before your marathon (the total distance covered during this period). Zone 1 includes any pace 10% or slower than your marathon pace.

About a week before your period, your breathing quietly speeds up. Most people never notice, but the pattern is remarkably consistent across thousands of cycles in our cohort of 1,882 women. Breathing rate deviation starts climbing around day minus seven, peaks just before period onset, and then returns to baseline within a few days. It is one of the most reliable physiological signatures of the menstrual cycle, and it is almost entirely invisible without a sensor.

What makes this finding more interesting is the PMS split. If you are someone who reports PMS symptoms, the breathing rate spike is noticeably sharper and taller. And after your period starts, instead of settling back to normal like it does for non-PMS users, it stays erratic for days. The body is signaling something is off well before you consciously feel it. This has practical implications: breathing rate could serve as an early, objective warning system for PMS onset, something no subjective symptom tracker can reliably do.

We looked at almost 3 million nights of sleep data from 1,882 women and tested which biomarker best tracks the menstrual cycle. Heart rate, HRV, and sleep all show the cycle to some degree, but the signal fades after a few days.

Temperature is different. It follows a clean 28-day wave that repeats month after month. The signal was so strong we could actually distinguish biological sex from temperature data alone, without anyone self-reporting. If you want to build cycle tracking into a health product, temperature is the metric that matters.