Wearable Score Conflicts: Push, Modify, or Recover?

One wearable says you are ready. Another says to slow down. Apple shows unusual overnight vitals while your training load looks ordinary.

This is not necessarily a malfunction. The products measure different inputs, use different time windows, and answer different questions. The goal is not to find the “true” color. It is to choose a training dose using the information underneath it.

What Each Product Currently Scores

These features change over time, so older descriptions can be misleading.

Oura Readiness

Oura currently says Readiness uses nine contributors:

1. Resting heart rate
2. HRV Balance
3. Body Temperature
4. Recovery Index
5. Sleep
6. Sleep Balance
7. Sleep Regularity
8. Previous Day Activity
9. Activity Balance

The contributors compare several short- and longer-term windows with personal averages. For example, Oura says HRV Balance compares the past 14 days with a three-month average, while Activity Balance compares 14-day activity with longer-term activity.¹

WHOOP Recovery

WHOOP currently describes Recovery as using HRV, resting heart rate, respiratory rate, sleep hours versus estimated need, sleep stages or quality, skin temperature, SpO2, and menstrual-cycle phase where applicable. It personalizes the calculation against the member’s baseline.²

That is broader than older summaries that listed only HRV, resting heart rate, respiratory rate, and sleep.

Garmin Training Readiness

Garmin Training Readiness is updated throughout the day. Current Garmin manuals list last night’s sleep score, recovery time, HRV status, acute load, sleep history from the last three nights, and stress history from the last three days.³

Garmin’s “acute load” is a branded input based on recent recorded exercise. It should not be converted into a universal injury-risk cutoff.

Apple Vitals and Training Load

Apple does not present a single readiness score. Apple Vitals shows overnight heart rate, respiratory rate, wrist temperature, blood oxygen where available, and sleep duration against typical ranges. A notification can appear when multiple metrics are outside those ranges.⁴

Apple Training Load is separate. It compares the intensity and duration of workouts from the last seven days with the previous 28 days and classifies load from well below to well above.⁵ That comparison describes recent relative workload; it is not an illness diagnosis or a complete recovery verdict.

Why the Scores Conflict

Different inputs

A low Oura score may be driven by body temperature or sleep regularity. A low Garmin score may reflect recovery time and recent load. WHOOP may place different weight on HRV or sleep. Apple may show a Vitals outlier without combining it with Training Load at all.

Different windows

One product may emphasize last night, another a rolling balance, and another the relationship between recent and prior training. A hard session can therefore affect each system on a different day.

Different measurements

Even when labels match, the measurement may not. Devices can sample HRV at different times and use different metrics or artifact handling. In a 2025 overnight validation, Oura Gen 4, WHOOP 4.0, and Garmin Fenix 6 showed different agreement with ECG in 13 healthy adults.⁶

Different personal baselines

Algorithms need time and consistent wear to learn a baseline. A new device, missed nights, travel, firmware changes, poor sensor contact, or a different wrist can temporarily create disagreement.

Proprietary weighting

The final formulas are vendor-specific. A study validating the accuracy of one sensor input does not by itself validate the composite decision produced from it.

A Conflict-Resolution Framework

1. Put symptoms and safety first

Do not use a high score to override fever, chest pain, unusual shortness of breath, palpitations, dizziness, fainting, or significant illness. Stop training and seek appropriate medical care for concerning symptoms.⁷

A wearable does not diagnose illness or clear you for exercise.

2. Open the score and identify the driver

Look past the color:

• Which contributor changed?
• Is the underlying measurement plausible?
• Was the device worn normally?
• Is the value unusual for your own history?
• Does a second device measure the same thing, or only something with a similar label?

Two scores can disagree while their raw inputs largely agree because the products weight them differently.

3. Add subjective and performance context

Consider soreness, motivation, pain, mood, perceived exertion, recent workout quality, and whether familiar work has become unusually hard. Subjective reports can capture information no wearable directly measures.⁸

Also check alcohol, travel, altitude, heat, hydration, menstrual-cycle context, medication changes, psychological stress, and an unusual sleep schedule.

4. Judge the pattern, not one morning

A single outlier deserves inspection. A persistent pattern across physiology, sleep, performance, symptoms, and recent training deserves more weight. There is no universal number of bad metrics that mandates a specific deload.

Avoid turning Apple’s seven-versus-28-day display, Garmin acute load, or any other vendor window into a homemade injury-risk algorithm.

5. Use movement as feedback only when safe

If you are symptom-free and the conflict is mild, start the warm-up easier than planned. Observe breathing, coordination, pain, heart-rate response, and perceived effort.

Proceed when the response is ordinary for you. Modify or stop if effort is abnormally high, technique worsens, pain appears, or you feel progressively worse. A warm-up is not medical clearance.

Push, Modify, or Recover

Push

The planned workout may be reasonable when you are symptom-free, underlying signals are close to their usual trend, recent performance is stable, and the warm-up feels normal. Document the session response so tomorrow’s decision has better context.

Modify

Choose a lower-cost version when evidence is mixed. That might mean reducing volume, lowering intensity, extending rests, replacing complex work with simpler practice, or choosing easy aerobic movement. The adjustment should match the session and the reason for caution rather than a fixed percentage.

Recover

Rest or choose very easy movement when symptoms, significant pain, marked fatigue, worsening performance, or a broad recovery decline makes the planned session a poor trade. Fever or concerning cardiopulmonary symptoms require a health response, not a score calculation.

How SensAI Uses Wearable Context

SensAI does not claim direct access to every vendor’s proprietary score or algorithm. Apple Watch connects through HealthKit. Garmin, Oura, and WHOOP data can flow through HealthKit when those services write the relevant fields there, and available metrics depend on permissions and vendor support.

The LLM coaching layer combines available aggregated recovery metrics with goals, equipment, schedule, constraints, planned-versus-performed workouts, and actual performance. It can generate a daily recovery summary and regenerate the weekly program from that context.

During a workout, you can request a change through quick actions or natural-language chat. These are user-requested modifications; a vendor score does not silently rewrite the session.

Raw HealthKit data stays on your device. Aggregated recovery metrics, sleep quality, and workout summaries used for coaching are sent server-side, and SensAI does not sell your data to third parties.

The Bottom Line

Readiness scores conflict because they are different products, not independent judges of one universal state.

Start with safety. Inspect the underlying contributor, compare it with your own trend, add subjective and performance context, and choose the smallest sensible change. The most useful answer may be push, modify, or recover—but it should come from the complete context, not whichever app used the most confident color.

---

References

Footnotes

1. Oura. “Readiness Contributors.” Oura Member Care, updated 2026. https://support.ouraring.com/hc/en-us/articles/360057791533-Readiness-Contributors ↩

2. WHOOP. “WHOOP Recovery.” WHOOP Support, updated 2025. https://support.whoop.com/s/article/WHOOP-Recovery?language=en_US ↩

3. Garmin. “Training Readiness.” Forerunner 570 Series Watch Owner’s Manual, updated 2026. https://www8.garmin.com/manuals/webhelp/GUID-25E3235D-44D2-4384-A591-DD1D71BEBCB1/EN-US/GUID-C21BE0C8-A08E-4DA1-B6C6-2E0E2DDDB372.html ↩

4. Apple. “Track Your Overnight Vitals With Apple Watch.” Apple Support, 2025. https://support.apple.com/en-us/120142 ↩

5. Apple. “Track Your Training Load on Apple Watch.” Apple Watch User Guide, 2026. https://support.apple.com/en-sg/guide/watch/apde4c07a6cf/watchos ↩

6. Dial MB, Hollander ME, Vatne EA, et al. “Validation of Nocturnal Resting Heart Rate and Heart Rate Variability in Consumer Wearables.” Physiological Reports, 2025. https://pubmed.ncbi.nlm.nih.gov/40834291/ ↩

7. Schwellnus M, Adami PE, Bougault V, et al. “International Olympic Committee (IOC) Consensus Statement on Acute Respiratory Illness in Athletes Part 1: Acute Respiratory Infections.” British Journal of Sports Medicine, 2022. https://pubmed.ncbi.nlm.nih.gov/35863871/ ↩

8. Saw AE, Main LC, Gastin PB. “Monitoring the Athlete Training Response: Subjective Self-Reported Measures Trump Commonly Used Objective Measures.” British Journal of Sports Medicine, 2016. https://pubmed.ncbi.nlm.nih.gov/26889611/ ↩