Where you sleep determines how much you sleep

At Terra Research, we harness wearable insights from devices like Fitbit, Apple Watch, and Oura to forge meaningful connections between daily activities and long-term wellness. This week we explored our latest anonymised dataset: 8,298 sleep records from October 25, 2025. 

By linking each user's last known exercise location to their subsequent sleep session, we approximate geographic influences on rest, acknowledging potential inaccuracies from travel but leveraging it as a robust proxy for nightly patterns. As always, ideally, we would have the exact location of the sleep event, but to do that, we would be sacrificing the size and the privacy of the dataset.

This analysis reveals how borders shape bedtime, with implications for personalised health models. We'll spotlight top countries, dive into average durations and variabilities, and highlight intriguing weekday-weekend shifts. We are looking at population-level statistics here, and as we’ve discussed in previous blogs, sleep is impacted by a wide range of variables.

Dataset Overview: A Worldwide Night's Rest

Our October 2025 snapshot captures a diverse user base, primarily from North America and Europe, reflecting Terra's footprint. We used the geolocation of the last exercise session of that day and presume that's the same location of that night’s sleep. While not infallible (e.g., post-workout flights could offset accuracy), it's a practical lens for global trends for a large dataset.

The key metrics include mean sleep duration (hours) and standard deviation (variability), aggregated by country. Overall, averages hover around 6-7 hours, aligning with health guidelines but underscoring opportunities for improvement.

Country-Level Sleep Metrics: Duration and Consistency

We computed means and standard deviations for all countries with notable records. Ireland tops the charts with over 7 hours, while emerging markets like Indonesia show shorter, more variable nights, potentially tied to urban hustle or equatorial light cycles. 

We don’t know how many users these records belong to; for countries with low values, they could come from only one user. Consequently, the mean may reflect the habits of only one user.  The table below shows the results for the 15 countries with the highest records.

Seasonal Considerations: Hemisphere Differences in October Sleep

Research indicates that populations tend to sleep less during summer months, with studies showing an average reduction of about 12 minutes compared to winter, often due to longer daylight hours and higher temperatures disrupting circadian rhythms. 

This seasonal effect can influence interpretations of our October data, particularly across hemispheres. In the Northern Hemisphere (e.g., US, UK, Germany), October marks autumn, with cooler weather and shorter days potentially promoting longer sleep—evident in higher averages, such as Ireland's 7.31 hours and Norway's 7.18 hours. 

Conversely, in the Southern Hemisphere (e.g., Australia, South Africa, New Zealand), October is spring, transitioning toward summer conditions that may begin to curtail sleep durations. Here, we see mixed but generally solid figures, such as New Zealand's leading 7.44 hours and Australia's 6.96 hours, though lower than some Northern counterparts. 

Equatorial regions like Indonesia (5.79 hours) show even shorter sleep, less affected by seasonal swings but impacted by consistent heat. These hemispheric nuances highlight the need for health models to incorporate seasonal adjustments for accurate global benchmarking.

We partitioned data into weekday (Sunday-Thursday) and weekend (Friday-Saturday) nights to detect "social jet lag." Globally, weekends extend sleep slightly, but nuances emerge:

• United States: A significant -0.26 hours drop on weekends, possibly from nightlife or screen time, informing Terra's fatigue risk models.
• Italy: A +0.51 hours uptick (non-significant). Is this Mediterranean leisurely weekend?

Research Relevance: Establishing Norms for Better Health Modeling

At Terra Research, analyses like this are foundational to our research mission. By mapping population-level sleep norms across geographies, we gain a deeper understanding of "typical" rest patterns and what constitutes a healthy baseline in diverse contexts. This allows us to benchmark individual users against these norms, flagging deviations that might signal emerging health issues. 

Moreover, it enables advanced phenotyping: clustering users into sleep "types" based on duration, variability, and cultural influences. We will continue to develop these phenotypes to power our machine learning models, from predicting recovery needs to tailoring interventions. Ultimately, it's about turning aggregated wearable data into personalised, proactive health insights.

In the future, we want to run a much larger location-based sleep study, with year-round records for each individual, to help us better understand genuine relationships. This provided a helpful first step to see if our model of using exercise location could work. 

[1] Mattingly, Scott M., Tushar Grover, Gabriel J. Martinez, et al. “The Effects of Seasons and Weather on Sleep Patterns Measured through Longitudinal Multimodal Sensing.” npj Digital Medicine, vol. 4, 2021, article 76, https://doi.org/10.1038/s41746-021-00435-2.