The programme was the starting point. The interesting work happened when the session did not match it.
I came to coaching through my own training. Since 2025 I have worked with one lifter, first using my own programmes and later plans developed with ChatGPT from the literature. Writing a plausible session was only part of that work. The decisions that followed—reducing work, clarifying an instruction, or keeping a technical issue open—deserve a closer look.
This essay follows those decisions through the available records. It is not a complete session-by-session diary. The athlete is unnamed, and the original correspondence is omitted.
The work changed over time.
Looking back, I can group the record into seven preparation blocks. Their names and most boundaries are retrospective. They help describe the changing task: establish a base, prepare for a meet, rebuild a movement, and make the workload manageable.
PREPARATION, BLOCK BY BLOCK
Seven chapters. Different questions. Different decisions.
Retrospective groupings by publication date and content, not a reconstructed calendar of every completed session.
01.07.2026 — 12.09.202630 plan messages
Recovery, constraints, and an intensification block
Intent and prescription
Recovery sessions in early July 2026, reduced grip demands, straps and controlled heavy work. August 2026–September 2026 prescriptions move toward heavier loads and fewer reps.
Documented outcome
September 2026 reports include squat 160, bench 140 and deadlift 210 kg. Forearms reportedly feel better; bench stability and deadlift lockout remain open.
Coaching lesson
Periodization, not proof of new max strength. Unstable bench and a strapless 215 kg pull are planned for 14 September 2026; no outcome is in this archive.
Explicit cycle-week numbering begins on 14 May 2026. Earlier labels and boundaries are editorial. The future 14 September 2026 plan is counted as a message dated 12 September 2026, never as completed work.
A heavier bar needs context.
The summer 2026 deadlift reports form a neat upward line. Presented alone, it would be easy to call it a strength gain. But the prescriptions moved from fours to triples to doubles, and the final load had already been lifted in competition. What the graph shows clearly is a change in the work.
THE SHAPE OF A TRAINING BLOCK
Heavier bar. Different prescription.
Load rises as the work changes. This is periodization—not a curve of new personal records.
REPORTED BAR LOAD / KG25 JUL 2026 — 11 SEP 2026
Select a report to inspect its context. Lines connect selected observations; they do not reconstruct every session.
DEADLIFT / PRESCRIBED WORK ONLY
Volume comes down.
kg × reps
2,1003 × 425 Jul 2026
2,2204 × 303 Aug 2026
2,2804 × 310 Aug 2026
1,7553 × 317 Aug 2026
8202 × 231 Aug 2026
8402 × 207 Sep 2026
Selected base prescriptions; warm-ups, accessories and fallback options excluded. These are planned values, not verified completed tonnage. The ambiguous 24 August 2026 prescription is omitted.
WHAT THIS ACTUALLY SHOWS
Fours → triples → doubles.
The deadlift block shifts toward heavier, lower-volume work. The first and last selected prescriptions move from 12 work reps to 4. That explains why bar weight alone is a poor progress metric.
210 kgApril 2026 meet result
210 kgSeptember 2026 training load
Different contexts: competition versus prescribed doubles with straps. No new 1RM or estimated max is inferred.
Some mistakes were in the text. Others happened after it.
A contradictory stop rule and an unclear set count are defects in a prescription. Ignoring an existing RPE fallback is a different problem. I want to preserve that distinction when describing what improved: a clearer plan is useful, but it is not proof of a better outcome.
The following notes separate the earlier instruction, the later record and the conclusion I can defend. Expandable exchanges are labelled paraphrases, not reproduced messages.
AUDITING MY OWN DECISIONS
What changed in the plans. What still needs work.
“Prescription corrected” means the instruction changed. Whether it improved the athlete’s outcome is a separate question.
A1Volume adaptationPrescription corrected
Reducing the work became part of the plan
Before / original decision
22 May 2026: bench 100 kg for 5 × 4, close-grip 90 kg for 4 × 6, plus incline dumbbells and upper-body accessories.
After / later record
3 June 2026, with low readiness: bench 100 kg for 4 × 4, close-grip 80–85 kg for 2 × 5; triceps optional, dumbbell pressing removed. A 105 kg alternative is available after warm-up.
Assessment
The two base barbell prescriptions change from 44 to 26 reps and 4,160 to 2,400–2,450 kg of planned tonnage. Warm-ups and accessories excluded. The athlete reports 105 kg, so the base option is not claimed as completed.
For PowerliftME: Show the changed load, sets, exercises and reason.
A2Decision logicPrescription corrected
A contradictory stop rule was removed
Before / original decision
27 July 2026: stop if pain increases, but another branch says to do two lighter sets at 95 kg if pain increases after the first work set. The actions conflict.
After / later record
29 August 2026: end bench at any stage if pain occurs; do not reduce weight just to finish the session. September 2026 repeats an explicit stop condition.
Assessment
The text logic was corrected. Reported improvement by 12 September 2026 cannot be attributed to this change alone.
For PowerliftME: Detect stop-versus-continue instructions triggered by the same condition.
A3ClarityPartly corrected
Vague back-off instructions became explicit sets
Before / original decision
26 January 2026: a long weight ladder and an instruction to finish with the same weight if it feels heavy, without an explicit number of remaining sets. Coach and athlete interpret the volume differently.
After / later record
From May 2026, work sets are listed individually. Later plans also spell out sets and reps, although compact notation remains.
Assessment
A documented improvement in format, not complete standardisation. The 130 kg bench on 27 January 2026 involved assistance and is not an independent successful lift.
For PowerliftME: Store weight, sets and reps as separate fields; generate the summary from the same record.
A4Plan versus actualExecution issue
The effort cap existed. It was not followed.
Before / original decision
3 June 2026: squat 130 kg for 5 × 3, target RPE 7–8, cap 8.5. If heavy by set three, reduce remaining sets to 125 kg.
After / later record
5 June 2026 reports remain at 130. The coach rates effort 9–9.5 and says the load should have been reduced. The athlete replies that he completed it.
Assessment
The missing piece was not a fallback—it was already there. Later restrictions become more explicit, but consistent adherence is not established.
For PowerliftME: Record actual RPE, the prescribed threshold and the decision taken.
A5Exercise specificationClearer; partial outcome
The pause needed one unambiguous definition
Before / original decision
16 May 2026: the exercise is named “3s Paused Bench”, while the summary says “Spoto / paused bench”. Later feedback still questions pause duration.
After / later record
11 September 2026: three seconds down, one second on the chest, a continuous path, and a full reset after a zigzag.
Assessment
The instruction is more measurable. Leg drive is still criticised that day; clearer text is not a fully corrected movement.
For PowerliftME: Store pause location, duration and tempo consistently across every view.
A6Meet decisionsSource discrepancy
The opener needs a single agreed version
Before / original decision
The 9 April 2026 written plan opens bench at 145 kg. The later case clarification says the coach advised 140 and the athlete did not follow it.
After / later record
The competition record confirms a successful opener followed by two misses. The timing and form of the 140 kg advice are not documented in the export.
Assessment
This does not prove the original 145 prescription was inherently wrong. Preserve the latest agreed decision and why it changed. Injury and psychological context come from the clarification.
For PowerliftME: Version the attempt card, with coach agreement and readiness alternatives.
A7Next experimentPrescribed; untested
The next experiment has no result yet
Before / original decision
September 2026 feedback still flags bench instability and bar path. Deadlift lockout remains a topic; summer heavy work uses straps.
After / later record
A message dated 12 September 2026 prescribes unstable suspended-load bench and a strapless 215 kg single for 14 September 2026.
Assessment
Neither a completed 215 kg pull nor a successful bench intervention can be claimed. The exercise and grip conditions change; evaluate their effects separately.
For PowerliftME: Keep an experiment open until an observation is recorded.
The limits of this audit
High volume alone does not prove a programming error. A missing report does not prove a missed session. Later improvement does not isolate the effect of one exercise. This audit evaluates the recorded decisions and feedback.
Technical feedback takes longer.
Foot pressure, bench setup and deadlift lockout recur across the record. There are encouraging sessions, followed by similar comments later. That recurrence matters: it prevents a single good report from becoming a success story the record cannot support.
THE ATHLETE BEHIND THE LOADS
Capacity is one thing. Repeatability is another.
What is working
Competition strength
Competition records establish successful bench and deadlift performances. The coach describes the meet deadlift as having reserve, without quantifying it.
Response to technical feedback
The archive contains specific better sessions: bench stability on 2 March 2026, deadlift lockout on 5 April 2026, and squat execution on 11 September 2026.
What remains open
Bench setup and leg drive
Positive sessions coexist with recurring setup and bar-path criticism. A good day is not yet a repeatable skill.
Lockout, bracing, and following the cap
Deadlift knee lockout and squat control remain technical tasks. The June 2026 RPE episode also reveals a gap between a valid fallback rule and its execution.
FIVE TECHNICAL THREADS
What we tried. What the record supports.
Observations over time, not controlled experiments. More than one thing changed.
01
SQUAT / FOOT CONTACTLocal effect
Footwear helped locally.
Problem
Weight shifted to the heels and the left knee drifted inward.
Intervention
Tried removing soft shoes, stance changes and unilateral work; later experimented with heeled and flat shoes.
Observed outcome
The athlete and coach reported better stability in specific sessions. Knee movement later recurred. No single footwear change solved the entire block.
02
DEADLIFT / SETUP & LOCKOUTStill in progress
A workable pull. An unfinished lockout.
Problem
Conventional pulls repeatedly showed weak pre-tension and early hip rise.
Intervention
Rotated conventional and sumo, bands, slow lowering and pauses; later returned to sumo and dead-stop resets.
Observed outcome
The heavier deadlift load had already been established in competition. September 2026 work returned to that load within a new cycle; knee lockout still drew criticism. No isolated effect of stance change is established.
03
BENCH / SETUPNot yet consistent
Strength was there. Repeatability was not.
Problem
Scapular position, elbow movement and shifting feet disrupted stability.
Intervention
Floor press, pauses, bands, grip and footwear changes; later full-foot contact and a revised setup sequence.
Observed outcome
Stable sessions in March 2026 and improved leg drive in July 2026 did not transfer consistently. September 2026 feedback again flags zigzag bar path and weak leg drive.
04
SQUAT / BRACING & DEPTHOne improved session
From too high to too deep.
Problem
Depth, rushing between reps and loss of trunk tension recur. By September 2026, fear of missing depth leads to excessive depth.
Intervention
Paused squats, breathing and bracing cues. A lighter filmed depth check was suggested.
Observed outcome
The coach praises the 11 September 2026 160 kg squat, with residual shaking and excess depth. Completion of the proposed 100 kg filming check is not confirmed.
05
LOAD / FOREARMSPartial improvement
Less extra work. Better reported symptoms.
Problem
Forearm discomfort recurs in July 2026 and August 2026, affecting several lifts.
Intervention
Removed direct grip work, reduced accessories, used straps and clarified bench stop conditions.
Observed outcome
By 11–12 September 2026 the athlete reports feeling much better. This coincides with several adjustments; it does not establish a single cause or medical outcome.
THE EVIDENCE LAYER
A chat archive is not a lab.
An anonymised editorial case. Expandable chats are labelled paraphrases; original messages are not included.
1,557non-service messages
176long-form plan messages
18calendar months with records
The archive, month by month
plan messages
404 2025
505 2025
206 2025
1407 2025
1208 2025
909 2025
910 2025
911 2025
912 2025
1101 2026
702 2026
2003 2026
804 2026
1205 2026
1506 2026
1207 2026
1408 2026
409 2026
Posting frequency, not attendance. One message may contain several sessions.
Counting the archive
Two private exports; service messages excluded. The source period spans 504 days inclusive. Plan messages are posts from the programme channel longer than 200 characters. Cycle boundaries are retrospective editorial groupings.
Reading a graph
Load points are selected athlete captions manually matched to a lift. Sets and reps come from prescriptions and are labelled separately. A caption such as “3. 195” identifies set three, not three repetitions. No estimated 1RM is calculated.
Limits of the evidence
Videos were not analysed. Technical assessments are the coach’s observations. No complete performed-set log, competition squat or three-lift total is available. Workload charts use only explicit base prescriptions.
Interpreting change
Match a problem, a later instruction and subsequent feedback. Recurrence keeps an issue open. Self-reported symptoms are not a diagnosis. Meet context was provided separately from competition records.
The product questions are concrete.
These examples give PowerliftME a useful set of cases to test against. They are proposals to check against the current app, not a retrospective list of features already delivered.
FROM CASE STUDY TO PRODUCT
What this asks of PowerliftME.
Proposals · not implemented
First check what the app already covers. Then use these concrete examples to test the planner, training log and plan validation. Features alone do not guarantee better lifting outcomes.
01
Check the plan before delivery
P0
Validate missing reps, conflicting pause definitions and mutually exclusive actions.
Acceptance example: The 27 July 2026 stop-rule conflict is flagged; the 16 May 2026 summary cannot silently change the variation.
Audit reference: A2 · A3 · A5
02
Connect prescription to actual work
P0
Show planned and reported work separately, including effort, assistance and deviations.
Acceptance example: The 5 June 2026 example surfaces the RPE ceiling and the 125 kg fallback instead of suggesting progression.
Audit reference: A4
03
Explain every adaptation
P1
Show what changed and why: load, volume, or exercise removal.
Acceptance example: The 3 June 2026 view explains the removed sets and dumbbell work, not just a new number.
Audit reference: A1
04
Keep records in context
P1
Separate training, competition and future plans. Preserve equipment, assistance and versioned opener advice.
Acceptance example: April 2026’s 210 kg does not become a September 2026 PR. Planned 215 kg never enters completed achievements.
Audit reference: A6 · A7
05
Close the feedback loop
P1
Give each technical intervention an observable criterion and follow-up. Distinguish one good session from repeatable improvement.
Acceptance example: A repeated lockout comment keeps the task open. A newly prescribed exercise cannot close it without feedback.
Audit reference: A5 · A7
WORK IN PROGRESS
PowerliftME
An iPhone app shaped by coaching practice. Exercise illustrations remain unfinished before release; this case adds testable requirements for its programming logic.