Cover illustration for Uncovering the Hidden Variable in a Stalled Training Block

What the AI found

“Every skipped Saturday run over ten weeks followed a Thursday run with cadence over 172 spm, regardless of total mileage or recorded heart rate.”

Before

Garmin sync alerts ignored, four missed long runs

After

One 4-minute Friday synthesis that protects Saturdays

The same system, three states — real screens, not a screenshot

1Starting
Apple Notes
  • May 11: Sat 18k abandoned at 4k. Legs like lead.
  • May 25: Skipped long run again. Slept 8 hrs though?
  • June 08: Bailed at 6k. Pace felt impossible.
  • Need to check Garmin logs sometime this weekend
2Working
Claude

Prompt

Cross-reference these 10 weeks of CSV run data against the four failed Saturday runs listed in my notes. What metric or combination differentiates the Thursdays before failed Saturdays from the Thursdays before successful ones?

Cross-reference these 10 weeks of CSV run data against the four failed Saturday runs listed in my notes. What metric or combination differentiates the Thursdays before failed Saturdays from the Thursdays before successful ones?

AI

Every skipped Saturday run over ten weeks followed a Thursday run with cadence over 172 spm (averaging 176 spm), regardless of total mileage or heart rate. On weeks where your Saturday runs succeeded, Thursday cadence stayed strictly below 168 spm.
3Implemented
Apple Health

168 spm

Thurs cadence cap

92%

Sat completion rate

4 mins

Weekly review time

IndividualSetup in use

Uncovering the Hidden Variable in a Stalled Training Block

How an amateur runner isolated the subtle neuromuscular trigger sabotaging his long runs using standard exports and a single prompt.

A 38-year-old half-marathoner and product lead, Northern Europe

Tools used

The real tools used here — swap any for your own equivalent. Each links to how we’d set it up.

4 min readWellness & AI editorial
1

Before anything was set up

His kitchen counter held an expensive sports watch that buzzed every morning with recovery scores he had learned to disregard. Ten weeks into a half-marathon block, four of his essential Saturday long runs had been abandoned before the six-kilometre mark. He kept blaming work stress, Friday dinners, or poor sleep, yet his sleep scores were consistently green. The raw split files sat locked inside his fitness app, unread and unmanageable, while self-doubt quietly chipped away at his autumn race goal.

Apple Notes
  • May 11: Sat 18k abandoned at 4k. Legs like lead.
  • May 25: Skipped long run again. Slept 8 hrs though?
  • June 08: Bailed at 6k. Pace felt impossible.
  • Need to check Garmin logs sometime this weekend
2

Setup, doing its job

During a short setup session, he exported ten weeks of activity data as a simple CSV and pasted it alongside his sparse training notes. Instead of asking for general coaching advice, he directed the model to isolate differences between Thursdays preceding completed long runs and Thursdays preceding abandoned ones. Within thirty seconds, the model highlighted a variable he had never once checked: cadence spikes during short mid-week intervals that fatigued his calves without elevating his cardiovascular strain.

Claude

Prompt

Cross-reference these 10 weeks of CSV run data against the four failed Saturday runs listed in my notes. What metric or combination differentiates the Thursdays before failed Saturdays from the Thursdays before successful ones?

Cross-reference these 10 weeks of CSV run data against the four failed Saturday runs listed in my notes. What metric or combination differentiates the Thursdays before failed Saturdays from the Thursdays before successful ones?

AI

Every skipped Saturday run over ten weeks followed a Thursday run with cadence over 172 spm (averaging 176 spm), regardless of total mileage or heart rate. On weeks where your Saturday runs succeeded, Thursday cadence stayed strictly below 168 spm.
3

The finished system, running on its own

The system now runs quietly in the background without manual data entry. Every Friday at 5:00 PM, an automated shortcut exports the week's raw run metrics into a single plain text note. Claude reviews the cadence and stride files against Saturday target readiness in under four minutes. He no longer bargains with fatigue on weekend mornings; he simply caps Thursday leg turnover at 168 steps per minute and lines up rested for Saturday's distance block.

Apple Health

168 spm

Thurs cadence cap

92%

Sat completion rate

4 mins

Weekly review time

9 of 10 completed

Long run consistency

4 min review

Weekly log time

Garmin Connectraw telemetry export

Captures granular ground contact time, stride length, and cadence without subscription lock-in.

Claudeweekly pattern analysis

Parses multi-column CSV exports and detects multivariate trends across weeks without hallucinating numbers.

Apple Notesrunning log & rules

Frictionless single-document storage for weekly constraints and simple session notes.

These are the tools used in this story. Any can be swapped for an equivalent you already trust.

Read the breakdown

This story runs on Setup. The tools and prompts above are the real build — swap any tool for your own equivalent and follow the same steps.

Suggested for you

Based on what you've been reading — always learning.

See all →