In this article
The short answer: Recovery and readiness scores are not a standardized measurement like a blood test. Each one is a proprietary composite: the company chooses which inputs go in, which part of the night it samples, which HRV metric it calculates, how long a baseline it compares you against, and how it maps the result onto a scale. Change any one of those choices and the same night of physiology produces a different number. Two apps disagreeing is expected by design, not proof that either one is broken. The practical fix is to pick one score as your primary, judge it only against its own history, and look at the shared raw inputs (HRV and resting heart rate trends) when two scores point in different directions.
- 01Why Scores Differ
- 02What Goes In
- 03Five Places They Split
- 04The Misconception
- 05What To Do
- 06FAQ
- 07Key Takeaways
- 08References
Why Do Two Apps Give Different Recovery Scores?
A recovery or readiness score is a product decision as much as a measurement. There is no agreed definition of "recovered," no reference instrument that every company calibrates to, and no shared formula. Each company builds its own model from the signals its device can collect and the questions it wants the score to answer, then publishes a description of the contributors without publishing the exact weights.
A standardized measurement
- Defined the same way everywhere (for example, resting heart rate in beats per minute)
- Checked against a reference method such as an ECG
- Two good devices should land close to each other
A proprietary composite score
- Each company decides its own inputs, weights, and scale
- No reference instrument measures "readiness" directly
- Two well-built scores can disagree on the same night and both be working as designed
That is why the more useful question is not "which score is right" but "what is each score built to notice." If you have not read it yet, our guide to why a recovery score changes day to day covers how a single score moves; this article is about why two scores move differently.
What Goes Into the Major Scores?
Based on how each company describes its own score, the overlap is real (HRV, resting heart rate, and sleep appear everywhere) but the supporting cast differs. Each design reflects a sensible choice about what a readiness number should capture.
Contributors each company describes
Oura Readiness
Described contributors include resting heart rate, HRV balance (recent HRV against a longer personal average), body temperature, recovery index (how early in the night heart rate settles), last night's sleep, sleep balance, previous day activity, and activity balance. Scored 0 to 100.
WHOOP Recovery
Built mainly from overnight HRV, resting heart rate, sleep, and respiratory rate. Shown as a percentage in three color bands: green (67 to 100), yellow (34 to 66), and red (0 to 33).
Garmin Training Readiness
Combines last night's sleep score, recovery time, HRV status, acute training load, sleep history over the previous three nights, and stress history over the previous three days. Scored 1 to 100 across five levels from Poor to Prime, and it updates through the day.
Notice what follows from those lists alone. A score that includes recent training load will drop after a hard block even if your overnight HRV looks fine. A score built mainly on overnight physiology will not. Neither choice is a mistake; they are answering slightly different questions.
The Five Places Two Scores Split
Even when two scores share an input, they can process it differently. These are the five design choices that explain most disagreements.
Which inputs count
Training load, temperature, respiratory rate, activity, and stress history are in some scores and not others. A late hard workout or a warm night can move one score and leave another untouched.
Which part of the night is sampled
Some devices average HRV across the whole night, others weight a particular window of sleep. In one validation study of five wearables, researchers matched their ECG reference to one device's own measurement window, the first four hours of sleep, rather than the whole night.
Which HRV metric is calculated
RMSSD and SDNN are different statistics computed from the same heartbeats. They are not interchangeable, and norms built from different recording lengths should not be mixed. Apple Health, for example, stores HRV as SDNN.
How long a baseline you are compared to
Every score judges last night against your own history, but the length of that history differs. A short baseline reacts fast and adapts fast; a long one is steadier but slower to accept a genuine change in your fitness or routine.
How the result is mapped to a scale
A percentage with three bands and a 1 to 100 score with five levels cut the same underlying change at different points. A 70 on one scale and a 70 on another do not describe the same state.
Raw sensor accuracy adds a smaller, separate layer. Independent validation work has found that consumer devices track overnight resting heart rate and HRV against an ECG reference with varying levels of agreement, so even the shared inputs are not identical numbers before any scoring begins. That is one more reason HRV is best read as a trend against your own baseline rather than as a single absolute value.
The Biggest Misconception
Common misconception
If my ring says 82 and my watch says 45, one of them is broken, and I need to find out which.
Disagreement is usually the expected result of different design choices applied to the same body. A score that weighs yesterday's training load and a score that weighs only overnight physiology should disagree the morning after a big session. Each one is telling you something true about the question it was built to answer.
Do not average the two
Adding two scores and dividing by two mixes different scales, inputs, and baselines into a number that no model produced. It hides the very disagreement that might be informative.
What Should You Do When Two Scores Disagree?
Pick one primary score
Choose the score whose inputs best match what you care about (overnight recovery, training load, or both) and make it your default daily signal. Use the second device for its unique signals, not as a second vote.
Judge each score only against its own history
A score means something relative to its own baseline. Compare today's number with that device's recent trend, never with the other device's number.
Drop down to the shared raw inputs
When the scores diverge, look at HRV and resting heart rate trends on both devices. If both show the same direction, that agreement is a stronger signal than either score alone.
Find the input only one score sees
If one score includes training load, temperature, or stress history and the other does not, that input often explains the gap. A hard session yesterday or a warm bedroom is a cause, not a contradiction.
Let how you feel break the tie
Research on athlete monitoring suggests self-reported wellness tracks training load well, and no single marker is strong enough to use alone. Your own check-in is a legitimate input, as covered in our guide on trusting a score versus how you feel.
For step 5 in more depth, see whether to trust a recovery score over how you feel. For step 3, the quickest way to see your own HRV trend in context is to know your personal range first.
Want a reference point for step 3? Work out your personal HRV baseline range:
Calculator
Find Your HRV Baseline
Copy them from your Oura, WHOOP, Garmin or Apple Health history, oldest to newest, separated by commas or spaces. Use one device only.
Frequently Asked Questions
Which recovery score is the most accurate?
There is no reference standard for "readiness" to test scores against, so accuracy in the usual sense does not apply to the composite score. Researchers can test the inputs, such as overnight heart rate and HRV against an ECG, but the final score is a modeling choice. The better question is which score's inputs match the decisions you want to make.
Why is my HRV a different number on each device?
Devices can calculate different HRV statistics (RMSSD or SDNN), sample different parts of the night, and filter noisy beats differently. Those differences change the number before any score is built on top of it, so compare each device's HRV with its own history.
Should I wear two devices to get a better answer?
Only if the second device adds a signal the first one lacks. Two devices measuring the same overnight heart data mostly give you two opinions on the same input. If you already wear two, pick one score as primary and use the other for its distinct features.
If one score is green and the other is red, should I train hard?
Check the raw HRV and resting heart rate trends first, then look for the input only one score includes, such as recent training load. If the gap is explained by a known cause, follow your primary score. If nothing explains it and you feel flat, ease off the top end of intensity for the day.
Can I switch devices without losing my recovery history?
Your old scores will not carry over in a comparable way, because the new score uses its own model and builds its own baseline. Expect a settling-in period of a few weeks while the new device learns your normal range, and avoid reading much into early scores.
What to remember
The short version.
- 1Recovery and readiness scores are proprietary composites, not a standardized measurement, so two apps disagreeing about the same night is expected by design.
- 2Scores split on five choices: which inputs count, which part of the night is sampled, which HRV metric is calculated, how long the baseline is, and how the result is scaled.
- 3A score that includes training load or stress history will react to things a score built only on overnight physiology does not see.
- 4RMSSD and SDNN are different HRV statistics and are not interchangeable, so raw HRV numbers also differ across devices.
- 5Pick one primary score, judge it only against its own history, and use shared HRV and resting heart rate trends to settle disagreements.
- 6Never average two scores. When both raw trends agree, treat that agreement as the stronger signal.
Keep reading
Related on Protocol
LearnRecovery
Should You Trust a Recovery or Readiness Score Over How You Feel?
When a score and your own check-in disagree, which one to weight, and what monitoring research says about each.
LearnRecovery
How to Read Your Recovery Score When HRV and Sleep Data Disagree
What happens inside a single score when its own inputs point in different directions.
LearnRecovery
How to Interpret Your HRV Data
Why HRV needs a personal baseline to mean anything, and how to read the trend rather than the night.
Protocol
One app. One coach. Everything connected.
Protocol brings your Oura, WHOOP, and Apple Health data into one place, reads the shared HRV and resting heart rate trends behind each score, and turns them into one clear action for the day instead of two competing numbers.
Get started freeReferencesSources
Key Studies
- Dial et al. (2025) Physiological Reports, volume 13, issue 16, e70527. Validated nocturnal resting heart rate and HRV from five consumer wearables against an ECG reference over 536 nights in 13 healthy adults, finding agreement varied across devices.
- Miller, Sargent, and Roach (2022) Sensors, volume 22, issue 16, article 6317. A laboratory validation of six wearable devices for sleep, heart rate, and HRV against polysomnography and ECG, showing devices are useful for sleep timing and duration but weaker on stage detail.
- Plews et al. (2013) Sports Medicine, volume 43, issue 9, pages 773 to 781. A review of HRV monitoring in elite endurance athletes that supports interpreting HRV as a trend against an individual baseline rather than as single-day values.
- Halson (2014) Sports Medicine, volume 44, supplement 2, pages 139 to 147. A review of training load monitoring noting that no single marker, subjective or objective, is strong enough to be used in isolation.
- Saw, Main, and Gastin (2016) British Journal of Sports Medicine, volume 50, issue 5, pages 281 to 291. A systematic review finding self-reported wellness measures tracked changes in training load with good sensitivity and consistency.
Key Reviews
- Shaffer and Ginsberg (2017) Frontiers in Public Health, volume 5, article 258. An overview of HRV metrics such as RMSSD and SDNN, explaining that they measure different things and that norms from different recording lengths are not interchangeable.
Apps and Tools
- Oura: Readiness Score Oura's own description of the Readiness contributors, including HRV balance, resting heart rate, body temperature, recovery index, sleep, and activity balance.
- WHOOP: How Recovery Works WHOOP's description of Recovery inputs (HRV, resting heart rate, sleep, respiratory rate) and its green, yellow, and red bands.
- Garmin: Training Readiness Garmin's owner's manual entry listing the six Training Readiness factors and the five score levels.