I've worn a sleep tracker for two years, which means I have a spreadsheet's worth of nights sitting in an export I never look at. For a lab project, I decided to stop looking at it as a spreadsheet and start looking at it as a place.
The build
Each night became a ring in a spiral, stacked by week, with color and height mapped to sleep stage — deep, light, REM, awake — and radius mapped to duration. Instead of a line chart, the whole two years became a structure you could walk into and orbit around, the way you'd walk around a physical model instead of scrolling through a report.
What the shape showed that the numbers didn't
The spiral made two things visible almost instantly that two years of dashboard-checking never surfaced. First, a visible flattening in the deep-sleep band that started almost exactly when I switched apartments — a correlation the daily numbers never made obvious because I was never looking at more than a week at a time. Second, a seasonal rhythm in total sleep duration that tracked daylight hours closely enough that it was clearly not a coincidence.
Neither finding required new data. Both were sitting in the export the whole time. What changed was the shape I put them in — spatial structure did the pattern-matching that a table of numbers was asking my eyes to do manually, and failing at.
Why this belongs in a company blog
This started as a personal curiosity project, but it's a clean small-scale example of the thesis behind everything else we build: data with structure — temporal, relational, hierarchical — usually has a spatial shape hiding in it, and finding that shape is often more revealing than any additional analysis you could run on the flat version. The same principle scales from two years of my sleep data to a year of a client's operational metrics.