The easiest analytics product to build is one that always produces an answer. The harder—and more useful—product knows when the available activity cannot support one.
A DECIMAL IS NOT CONFIDENCE
Calculation can create certainty the sensor never recorded.
A heart-rate metric needs credible heart-rate evidence. A terrain metric needs usable altitude. A durability reading needs enough continuous pace and heart-rate samples to compare the run across time.
If those inputs are weak, a polished number does not repair them. It only hides the uncertainty. Runome distinguishes measured fields, derived metrics and estimates so the result remains connected to its source.
One complete-looking result assembled from partial or unreliable evidence.
The required streams did not pass the checks for this metric.
THE EVIDENCE GATE
Different metrics require different proof.
Available metrics depend on what the activity recorded. Review the source, sensor coverage and activity conditions when a value is missing or an estimate needs more context.
Dropouts or invalid samples can remove HR-based economy, load and durability signals.
GPS drift, long pauses and missing streams can weaken pace-to-heart-rate relationships.
Elevation depends on the quality of the recorded altitude. Check the route profile before interpreting a climb or descent.
- 01
Enough duration
Very short runs may not reach a meaningful steady state, so confidence falls and some metrics are not calculated.
- 02
Enough coverage
A summary average cannot replace the continuous samples required for drift, decoupling or route-level analysis.
- 03
Acceptable context
Heat, hills, treadmill running and significant pauses can change what the metric is able to claim.
- 04
Consistent definitions
The same versioned calculation should mean the same thing wherever the value appears.
PARTIAL EVIDENCE STAYS PARTIAL
Run Quality is hidden when its foundation is incomplete.
Run Quality helps you review the activity as a whole. Its availability depends on the recorded evidence; useful individual measurements can remain available even when the overall view is missing.
Start with the values you can inspect: distance, duration, route and any available sensor recordings. An unavailable summary does not erase a completed run.
The summary record is supported and remains visible.
Not enough continuous economy plus stress or durability evidence.
Location samples pass the relevant checks.
Too few altitude samples with usable vertical accuracy.
Example availability states, not a user activity.
ABSENCE IS INFORMATION
“Not available” says something useful.
The workout can still be valuable even when a sensor-dependent metric is unavailable.
The recorded inputs cannot responsibly support this particular conclusion.
Data quality is not a score for the athlete. It is a limit on what the software should claim.
REPAIR THE EVIDENCE, NOT THE NUMBER
Start with the source.
If a metric is unexpectedly missing, check the heart-rate source, recording permissions and whether the activity imported with streams. Re-syncing the provider or recalculating after an import finishes can restore values when the evidence exists.
If the source data is genuinely incomplete, leave the gap intact. A trend built from fewer trustworthy values is more useful than one filled with guesses.
The most professional answer is not the one with the most decimals. It is the one that stops exactly where the evidence stops.