Names and aliases
match latency, time until first eligible match
Formula
matched_at - eligible_demand_at; report median and quantiles
Metric card
| Unit | time: seconds, minutes, hours, or days |
|---|---|
| Data source | Timestamped qualified-demand and first-eligible-match events using one documented timezone. |
| Frequency | By demand cohort, recalculated after the selected observation window closes. |
| Segments | micro-market, category, geography, price band, channel, cohort |
Interpretation errors
- Using only matched observations and ignoring right censoring.
- Reporting only the mean for a skewed distribution.
- Comparing platforms with different start events or match criteria.
Example
Illustrative example: durations of 10, 18, and 70 minutes have a median of 18 minutes. These values do not describe a real platform.
Use
Fix the eligible-entry event, first valid-match criterion, and maximum waiting window. Report the median, upper quantiles, and the share still unmatched when the window closes.
Anti-example
Do not claim a speed improvement if the longest unresolved observations disappeared from the calculation.
Limitations
- A fast match can be irrelevant or fail to become a transaction.
- The measure is sensitive to operating hours, seasonality, and window length.
Related metrics
fill-rate, unmet-demand-share, match-quality, time-to-first-transaction
Context sources
These works provide two-sided-market context; the formula above remains an author-defined working definition.
- Jean-Charles Rochet, Jean Tirole. Platform Competition in Two-Sided Markets (2003).
- Mark Armstrong. Competition in Two-Sided Markets (2006).