Recovery Signals and the 3-Point Coach Check

Recovery Signals and the 3-Point Coach Check

When fatigue rises, the next move is not automatic: fix food, change training, or wait for adaptation to show itself.

The Drive recovery guide says the scale is the wrong primary tracking tool during metabolic recovery, because body weight shifts from glycogen, water, digestive content, and hormonal fluid shifts; the correct markers are training performance, energy quality, sleep, mood stability, and recovery rate between sessions. That mechanism is fluid and substrate restoration, not sudden tissue gain. In practice, the falsifiable coaching thesis is simple: if those recovery markers are improving, do not chase the scale; if they are flat or worsening, the next change should match the bottleneck—nutrition if output and fullness are down, training if performance is lagging despite adequate recovery, and patience if the trend is already moving the right way.

Recovery signal quality matters more than single data points

AI coaching gets loudest when it acts like every check-in is a diagnosis. It is not. A weekly prompt, a wearable score, or a bodyweight trend can all be useful, but only if the coach understands what kind of signal is actually being measured.

On the recovery side, the key question is not “did weight go up?” It is “what changed in the system that would explain that weight?” The Drive guide lays out the issue clearly: early recovery and reverse dieting can add 3–7 lb in the first few weeks from glycogen restoration, hydration, digestive content, and hormonal water shifts. None of that is fat. So if an AI coach treats a fast scale jump as a problem, it will make the wrong adjustment for the wrong reason.

That is the first recovery-signal rule: the scale is a lagging, noisy proxy during recovery. It can be useful context, but it is a poor first-order decision tool when the athlete is depleted, flat, or coming out of a diet.

What the coach should watch instead

The better signals are the ones tied to function:

• training performance
• energy quality
• sleep
• mood stability
• recovery rate between sessions

These are not “nice to have” soft metrics. They are the closest thing to an operational readout of whether the athlete is tolerating the current stress load.

The Drive guide calls training performance the most reliable leading indicator of metabolic recovery. That is the kind of statement coaches can use without overcomplicating it: if lifts are going up or staying stable, workouts are being completed without feeling destroyed afterward, and the athlete is not accumulating obvious fatigue, then recovery is probably moving in the right direction even if scale weight is noisy.

This is exactly where AI can help or fail. A decent system can organize trend data and surface changes faster than a human memory can. A bad one can overfit to a single number and ignore the pattern. Coaches do not need a model that “knows” recovery in some mystical sense. They need one that recognizes when the signal quality is good enough to act and when it is not.

The next change depends on the bottleneck

If recovery markers are deteriorating, the next move depends on which part of the system is actually underperforming.

1) If fullness, pumps, and output are down: look at nutrition first

The Drive material points to glycogen as a major early recovery variable. That matters because a depleted athlete can look and feel “off” long before any true tissue change has occurred. If training is getting harder to complete, the athlete feels flat, and there is no obvious sign of adaptation, food is the first lever to examine.

That does not mean every bad session needs more carbs. It means the coach should ask whether the athlete is still in a low-energy state, underfed around training, or failing to restore enough substrate to support repeat performance.

For AI coaching, this is the right decision tree: when the recovery signal says “low fuel,” the next experiment is nutritional, not motivational. Increase food before you increase complexity.

2) If food is adequate but performance is sliding: look at training stress

Justin Harris’s coaching notes make the point bluntly in a bodybuilding context: pushing for a particular scale weight in the offseason is almost always a net negative to progress, and the scale has held back more bodybuilders than probably anything else. He also ties scale movement to water and fat over short time frames, not meaningful progress.

That same logic applies to training stress. If the athlete is eating well, bodyweight is stable or gradually rising, but sessions keep degrading, the issue may not be nutrition at all. It may be that the program is too aggressive for the current recovery state.

In AI terms, the model should not recommend more work just because the athlete is “not progressing.” It should ask whether the current load is still producing an adaptation signal or is simply adding fatigue. If the latter, adjust volume, intensity, exercise selection, or frequency before adding more food.

3) If markers are stable and only the scale is weird: wait

This is the hardest instruction for coaches who like action, and it is often the correct one.

Ciaran Lewis’s notes are a good example of restraint: “I’ll hold on changes” when the athlete looked bigger and fuller and conditioning still looked good, followed later by adding food when the athlete was “possibly a bit flat” but the metabolism was “booming.” That sequence is the right logic. Hold when the signal is mixed but moving well. Add food when the signal says under-recovered and the goal is restoration. Do not create a problem just because the scale or appearance is temporarily ambiguous.

Patience is not passive. It is a decision to let the current intervention finish its job when the available signals still support it.

Why AI tools need better recovery logic than motivational language

A lot of AI coaching software is built to sound responsive. But recovery decisions are not about sounding responsive. They are about correctly classifying noise.

A useful system should separate three cases:

1. Signal improving: keep the plan.
2. Signal worsening and fuel looks low: adjust nutrition.
3. Signal worsening and fuel looks adequate: adjust training.

That is a simpler, more defensible framework than trying to infer readiness from one wearable score or one scale reading. It also matches how experienced coaches already behave in practice: they watch the pattern, not the isolated datapoint.

The Drive guide explicitly recommends substituting biofeedback markers for scale dependency. The practical implication is not “ignore bodyweight forever.” It is “do not let bodyweight outrank function during recovery.” If the athlete is sleeping better, training better, and recovering faster between sessions, the trend is probably on track even if the mirror and scale are temporarily confusing.

What to tell coaches using AI this week

When the question is recovery signal quality, the right answer is usually not a dramatic intervention. It is a clean classification.

• Nutrition change when performance, fullness, and energy all point toward low substrate.
• Training change when food is adequate but fatigue is still accumulating.
• Patience when the athlete is trending better and the “bad” signal is mostly scale noise.

That is the real advantage of AI in coaching: not prediction theater, but better triage. The model should help a coach make the next change for the right reason. If it cannot do that, it is just another source of noise.

Sources Used

• wiki/drive-nutrition-recovery-tracking-and-biofeedback.md
• raw/kahunas-export/2026-05-28/clients/ciaran_lewis___members-xamwdoz3ggzeq1n-vdooo7iyblv8jm--hfkjzwbhrec.json
• raw/kahunas-export/2026-05-28/clients/michael_main___members-a2m88q4kyryqrsbdgta-x0mipybv-fzeobfolztzovk.json
• wiki/comprehensive-performance-nutrition-vol3.md