Not everyone logs like a spreadsheet wizard. My training 'log' for years was a single ever-growing note in my phone, full of fragments like 'bench 3x5 60 felt ok' and 'legs - knee weird, cut short' and the occasional cry for help. When I first wanted to run it through the Are You Actually Getting Stronger? Honest Read of Your Training Log prompt, I assumed my chaos was too messy to be useful. I was wrong, but I did have to learn how to feed it.
This is the under-discussed half of using analysis prompts: the output is only as good as the input you can paste in. So this post isn't about the insights — my other blog covers those — it's the practical workflow for turning a real, messy human log into something Grok 4.3 can actually reason over.
Step one: one line per session, consistently
The single highest-leverage change was forcing every session onto one consistent line: date, then each lift as 'name weight x sets x reps,' then a short note. So 'bench 3x5 60 felt ok' became '2026-04-02 | Bench 60kg 3x5 | OHP 35kg 3x6 | felt strong.' It took me twenty minutes to reformat three months of notes, and that twenty minutes was the difference between an analysis that could track my bench week over week and one that just shrugged.
- Date every session — without dates, the prompt can't compute your real completion rate or spot a pattern like missed Fridays
- Keep lift names identical every time — 'Bench' and 'bench press' and 'BP' read as three different lifts and break the trend analysis
- Always log the weight, even on a bad day — a skipped number becomes a blind spot in the trend
- Keep your honest notes ('knee weird,' 'rushed') — they're how the analysis explains a plateau, not just spots it
Step two: tell it your starting goals
The prompt has a starting_goals field and I used to leave it vague, which wasted half the value. The analysis grades you against your stated goals, so 'add 10kg to my squat and train four times a week' gives it a yardstick. When I finally filled it in properly, the output stopped being a generic trend report and started telling me whether I was on track for the specific thing I'd set out to do — which, on the four-times-a-week front, I emphatically was not.
Step three: run it on a real time window
Don't paste two weeks and expect wisdom. The honest read needs enough data to separate signal from a bad night's sleep — four to eight weeks is the sweet spot. The first time I ran a clean, dated, eight-week block through it, the difference was night and day from my early messy attempts. It correctly flagged that my 'consistent' training had a 68% completion rate and that my squat hadn't moved, because for the first time the data was legible enough for it to see clearly.
The trick that finally made me log in real time
Reformatting three months of old notes was a one-time chore. The real win was changing how I log going forward so I never have to do that cleanup again. What worked, embarrassingly, was lowering the bar. I used to think a 'proper' log meant a fancy app with charts and rest timers, so I'd download one, use it for a week, and quit. Now I just have one note titled 'LOG' pinned to the top of my notes app, and the rule is: before I leave the gym, I type one line in the format the prompt likes. Ten seconds. No app, no friction, no streak to break. Because the format is already analysis-ready, the end-of-block read takes zero prep — I copy the whole note, paste it in, done.
- Pin one plain note titled 'LOG' — no app, no charts, nothing to abandon after a week
- Type the line BEFORE you leave the gym, while the numbers are in your head, not from memory that night
- Use the exact same line format every time so the note is paste-ready when the block ends
- If you miss a session, log a dated 'MISSED — reason' line; the prompt uses the gaps, so don't hide them
One more practical note on the model side: because Grok 4.3 is reasoning over potentially dozens of dated lines, the cleaner and more uniform your formatting, the less it has to guess, and the sharper the trend analysis gets. Messy input doesn't break it outright, but every inconsistency — a renamed lift, a missing date — is a small fog the model has to see through. Tidy input, clear read. It really is that direct a relationship.
The meta-lesson stuck with me: a good analysis prompt rewards good logging, and good logging is a tiny habit. I now spend ten seconds after every session writing one clean line, specifically so that the monthly read is sharp. If your log is a mess, don't let that stop you — spend the twenty minutes, reformat it, and grab the Are You Actually Getting Stronger? Honest Read of Your Training Log prompt on Prompt Dock. The insight is waiting in your notes app. You just have to make it readable. And once you've got the read, my Weekly Training Block From Your Goal, Gear, and Real Schedule prompt is where I turn those findings into the next block.