Period analysis
A practical guide to the weekly training review — what the AI Assistant analyses across a full training period, what it surfaces that single-session analysis cannot, and how to use it most effectively.
What the analysis produces
Period analysis reviews a defined date range and returns a structured report with up to seven sections:
| Section | What it contains |
|---|---|
| Week highlight | The single most important takeaway from the period, with specific metrics |
| Training summary | Volume completed, workout types, and how the week aligns with the current training phase |
| Health & recovery | HRV, resting HR, sleep quality and duration, fatigue level — with actual values and trends |
| Training load analysis | Acute and chronic load, load ratio, week-over-week progression, and 4-week pattern |
| Performance insights | Pace trends, HR behaviour, cardiac drift, aerobic efficiency — only included when meaningful data exists |
| Areas of concern | Overtraining signals, volume progression violations, recovery deficits — only included when actual concerns exist |
| Recommendations | Specific, measurable guidance for the next week: target volume, session structure, HR zones, recovery protocols |
Performance insights and areas of concern are omitted when there is nothing meaningful to say. The report does not fill space with generic observations.
What the system fetches automatically
The AI Assistant retrieves a 21-day historical window anchored to the end of the analysis period. You do not need to provide any data manually.
Training load metrics:
- Acute load (5-day average in AU) and chronic load (21-day average in AU)
- Load ratio (relative fatigue, rTFF) — optimal range is 8–13
- Weekly volumes for the current week, last week, and week before
- Last hard workout date and consecutive hard training days
Health and recovery data:
- HRV — individual daily entries for trend analysis against 7-day baseline
- Resting HR — with 7-day baseline for comparison
- Sleep duration, consistency, and quality scores
- Fatigue, illness, and pain levels (each on a 0–4 scale)
- Stress level (0–100)
Training history:
- All workouts completed during the period, with duration, load, and any athlete comments
- Weekly statistics for the last 4 weeks: total duration, workout count, hard session count
- Volume trend direction and rest days
Athlete profile:
- Age, gender, weight
- HR zones, thresholds, and FTP if configured in Good Coach App
- Recent 10K performance as a fitness reference point
Training phase context:
- Current training phase (base, build, peak, recovery, transition) — inferred from the active mesocycle in Good Coach App
- Upcoming competitions and goals with dates
- Coach memory — your methodology and preferences, which shapes how the AI Assistant frames the analysis and recommendations
- Athlete memory — the accumulated profile of this athlete built from previous analyses
Your inputs
Athlete and date range (required): The athlete to analyse and the period start and end dates. The date range does not need to start on a Monday — any range is accepted.
Coach notes (optional): Context or focus areas that steer the analysis. Treated as high-priority instructions. Use these to surface things the data cannot capture — a stressful week at work, a change in goals, a race result that affected motivation, or a specific pattern you want the AI Assistant to investigate.
Overtraining risk assessment
The AI Assistant automatically evaluates five signals and reports how many are present:
| Signal | Threshold |
|---|---|
| HRV decline | >20% below 7-day baseline for 3+ consecutive days |
| Elevated resting HR | >5 bpm above 7-day baseline for 3+ consecutive days |
| Poor sleep | Average <7 hours per night over 7 days |
| High load ratio | rTFF >15 |
| High subjective fatigue | Fatigue score ≥3 on the 0–4 scale |
Risk levels:
- 1 signal — low risk; monitor but no immediate action needed
- 2 signals — moderate risk; recommend load reduction and close monitoring
- 3 or more signals — high risk; immediate intervention required
When the areas of concern section appears in the report, it always quantifies the signal count (e.g., "2/5 overtraining signals present") so the severity is clear at a glance.
Training load interpretation
The load ratio (rTFF) is the ratio of the athlete's acute 5-day average to their chronic 21-day average. It captures whether current training stress is in proportion to the athlete's established baseline.
| Load ratio | Interpretation |
|---|---|
| Below 8 | Training load is below the athlete's baseline — possible detraining if sustained |
| 8–13 | Optimal range — current stress is proportionate to chronic load |
| Above 13 | Load is rising faster than the baseline — monitor for accumulating fatigue |
| Above 15 | Overreaching risk — one of the five overtraining signals is active |
A high load ratio is not always a problem. A planned loading week in a build phase should produce a temporarily elevated ratio. Context from the training phase and health data determines whether it needs intervention.
The AI Assistant also checks the 10% progression rule. A weekly volume increase above 10% is flagged, above 15% is a high-priority concern, and the specific percentage is always stated — not just described as "significant" or "large".
How phase affects the analysis
The training phase is inferred from the athlete's active mesocycle in Good Coach App. It shapes what the AI Assistant considers appropriate for the week.
| Phase | Expected intensity mix | What the AI Assistant looks for |
|---|---|---|
| Base | 70–80% easy | Aerobic base building, volume consistency, no excessive hard sessions |
| Build | 60–70% easy | Progressive load, 2–3 hard sessions, threshold and interval work |
| Peak | High intensity maintained, volume reduced | Quality preserved, volume dropped 20–40%, signs of freshness |
| Recovery | 90%+ easy | Full compliance with easy-only training, adequate rest |
A mismatch between the phase and actual training is flagged. If the athlete is in a base phase but running four hard sessions per week, the AI Assistant will note it.
If the athlete has no active mesocycle, the analysis defaults to base phase assumptions.
Coach notes — what makes them useful
Period analysis synthesises a week of data. Coach notes are most valuable when they provide context that data cannot carry.
| Less useful | More useful |
|---|---|
| "Check training" | "Athlete reported heavy legs all week — does load data explain it?" |
| "Volume was low" | "Athlete travelled Monday to Wednesday — planned reduced week" |
| "Hard week" | "This was the peak volume week of the 4-week block — assess accordingly" |
| "Race coming up" | "Athlete races on Sunday — flag anything that suggests they are not ready" |
The AI Assistant will not invent context. If something important happened during the week that is not in the data, it will not know unless you say so.
Memory and continuity
See AI Assistant memory for a full explanation of how both memory types work across all generators.
After every period analysis, athlete memory is automatically updated with observations worth retaining — training patterns, how the athlete responds to loading, health trends, effort calibration tendencies. This is the primary way athlete memory grows: each period analysis adds to the cumulative picture.
Coach memory is not updated from period analysis. It is built from week generation and cycles generation.
A new athlete with no prior analyses will have no athlete memory yet. The AI Assistant will work from profile data and your notes, and memory will begin building from the first analysis you run.
What the analysis cannot do
It cannot tell you what to plan next week. The recommendations section gives specific guidance, but the AI Assistant does not know what you had in mind, what the broader block looks like, or what matters to this athlete beyond the data. Use the recommendations as a starting point, not a prescription.
It cannot account for context outside the data. External stress, sleep disruption from causes unrelated to training, travel fatigue, illness that the athlete did not log — none of these are visible unless the athlete logs them or you note them. If you know something happened during the week, say it in coach notes.
It will not prescribe race-specific training. For upcoming competitions within 3 weeks, the AI Assistant flags the timeline and recommends coach review. It does not prescribe tapers, race-week protocols, or race-type-specific workouts — those decisions belong with the coach.
It analyses the period you give it, not a longer arc. For 4-week block-level pattern recognition, the AI Assistant uses the weekly statistics from the previous weeks as context — but if you want a deeper multi-week review, run the analysis across a longer date range.