What a Reaction Percentile Means
A percentile indicates your relative ranking against a benchmark distribution, not your absolute physiological limit.
Percentile context is useful for comparison, but quality depends on session integrity and cohort assumptions.
- Higher percentile means faster relative performance.
- Median-based percentiles are more stable than best-click percentiles.
- Always read percentile together with consistency score.
Why Cohort Adjustments Matter
Device class and age band can shift expected ranges, so percentile mapping should not assume one universal baseline.
Desktop/mobile differences are often large enough to distort direct ranking without adjustment.
- Use device cohort for fair interpretation.
- Use age band for context, not deterministic labeling.
- Keep the same cohort assumptions when tracking progress.
Integrity Checks Before Percentile Trust
If integrity flags are high, percentile outputs may be less reliable for decision-making.
Early-click bias and suspicious low repeats can create misleading rank jumps.
- Check early-click rate first.
- Inspect suspicious-repeat warnings.
- Re-run with warmup and controlled setup.
Best Practice for Percentile Tracking
Track weekly median percentile in one mode before changing training parameters.
Use challenge links for peer comparison only after stable baseline sessions.
When percentile changes look large, compare the underlying milliseconds and integrity notes before treating the movement as real progress. A cleaner median with fewer warnings is stronger evidence than a single high rank from a noisy session.
Why Percentiles Help More Than Raw Milliseconds Alone
A raw time tells you how fast one session was. A percentile gives that number social context by showing roughly where it sits inside a broader benchmark frame.
That context is useful when the user needs a decision, such as whether a result looks typical, above average, or worth retesting before drawing conclusions.
- Use percentiles to explain the result to non-specialists.
- Keep raw milliseconds visible so the benchmark stays concrete.
- Use both numbers together when you compare sessions.
What Percentiles Do Not Measure
A percentile cannot tell you whether the session was clean, whether hardware latency was stable, or whether the result came from a trustworthy routine. It only adds ranking context to the captured number.
That is why percentiles should sit beside integrity checks and consistency signals rather than replacing them.
- Do not use percentile rank as a substitute for session quality.
- Check consistency before you compare percentile jumps.
- Retest when the benchmark conditions are weaker than usual.