Field Notes from a Scale • A forensic read of your Google Fit export

The Ruler‑Straight Disappearing Act

Most weight loss stories look like weather: messy, moody, full of little storms. Yours looks like architecture. And that’s the mystery.

1) The first clue is a gap

The dataset spans . But the scale only speaks on of those days. A detective’s instinct kicks in: missingness is a motive.

Wait, really? Most of your weight data is concentrated in 2025. That’s not a footnote. It’s the plot.

The chart begins wide—years at a glance—because stories don’t start where they get interesting. They start where you can still pretend nothing happened.

2) Zoom in: 2025 is when the case opens

On , the scale reads . By , it reads . That’s in .

Total change across 2025 weigh‑ins (Jan → Dec). But the shape matters more than the total.

The dots are daily weigh‑ins. The line is a 7‑day trend—because bodies are noisy instruments, and detectives don’t convict on a single witness.

3) The curve isn’t one story. It’s three.

A piecewise fit (think: “where does the ruler bend?”) finds two breaks: and .

Before the first break, weight falls like a trapdoor. After it, the drop continues—just slower. And then, after the second break, something even stranger happens: the line stops being a slide… and becomes a tightrope.

Wait, really? The steepest loss happens before your highest activity weeks. The miles aren’t the main character.

4) Milestones: five numbers, five scenes

Stories need scenes you can point to. These are yours: the first days you crossed 80, 75, 70, 65, and 63 kg. The chart marks them like thumbtacks on a corkboard.

The temptation is to celebrate the milestones. The smarter move is to ask: what changed right before them? That’s where leverage hides.

5) The smoking gun: the fastest fortnight

The most dramatic 14‑day stretch ends on : roughly in two weeks.

That kind of drop is rarely “fat alone.” It’s usually a cocktail: glycogen, water, gut content—plus whatever real deficit you created. The point isn’t to discount it. The point is to not mistake it for a permanent law of physics.

Fastest 28‑day window (ends ). Useful as a clue, not a promise.

6) Suspects: steps, “calories,” and the vanishing correlation

Here’s the detective trap: you see a clean weight curve and you assume one clean cause. The data argues back.

Tap a suspect. The scatterplot shows weekly weight change vs that metric. Correlation is not causation—yes. But lack of correlation is a kind of evidence too.

Wait, really? Weekly steps barely track weekly loss. Burned calories and heart minutes track better. That hints at intensity (and diet) being more decisive than sheer volume.

7) The underrated feat: maintenance

Most people can lose weight for a month. Fewer can keep it off once life gets bored. After , your weight hovers in a narrow band.

The chart draws a maintenance envelope—because once you’re in the endgame, the metric changes: it’s not “how fast can I drop?” It’s “how stable can I stay?”

So what? Build rules for trend exits, not day‑to‑day fluctuations. A scale should be instrumentation, not judgement.

8) A late clue: body fat % arrives after the plot twist

Body fat % appears only for days (Nov–Dec 2025). That’s too late to explain the cut. But it’s perfect for questioning the maintenance story: are you stable because you’re “done,” or because you built a system?

The body‑fat line is deliberately treated with suspicion: consumer BIA readings are sensitive to hydration and timing. So the chart emphasizes trend, not drama.

Open the evidence drawer (methods • coverage • caveats • downloads)
Days in export
From Google Fit daily metrics
Days with weight
Dense in 2025; sparse before
2025 weigh‑in coverage
Days in 2025 with weight logged
Days with body fat %
Nov–Dec 2025 only

Method (changepoints): piecewise linear fit on 2025 daily weight series, selecting two changepoints that minimize total squared error with a minimum segment size. This is descriptive: it finds “where the story bends.”

Critical caveats:
• The “Calories (kcal)” field here is what Fit exported for energy expended, not food intake.
• Body fat % from consumer BIA devices can swing with hydration and timing; treat single-day values as noisy.
• Correlations are exploratory; they don’t prove causality.

External sanity checks (non‑data links): CDC and NHS both suggest gradual loss (often around 0.5–1 kg/week) as a common “sustainable” target. (Links: CDC, NHS)

Download embedded daily data (JSON)
Download derived weekly data (JSON)
Download current chart (SVG)
This page is self-contained. Data is embedded in the HTML. No cookies, no trackers, no server.
Built as a narrative audit: the goal is not to flatter the curve, but to explain it. If you later add diet logs, training sessions, travel markers, or stress/sleep, this scaffolding can absorb them.
Weight timeline
7‑day trend
All years
Markers
Hover for details. Scroll the story to reveal layers.