Quality
Lag
Also called reporting lag.
Lag is the delay between a hit occurring and the report reflecting it. It can be seconds, minutes or hours.
How it is measured
Fire a test visit and note when it appears in each view. The time it takes is the lag for that view.
Tarsier's Live view shows hits within moments. Historical reports take longer, and the last few minutes can be revised.
Worked example
A flash sale team watches the dashboard for an order spike. At 12:05 the chart is flat, so they cut the discount. The orders arrive in the report at 12:19. The shop sold 41 units at the lower price that it would have sold at the higher one.
Next time they set a rule: no decisions on a figure less than 30 minutes old.
How it differs
Lag is the delay itself. Data freshness describes how recent a report is. Processing latency is the portion spent inside the pipeline. Real-time analytics shortens it.
Common errors
Acting on an incomplete window. Assuming all views have the same lag. Not testing. Ignoring batched sources. Forgetting time zones.
In practice
Fire a test hit now and time it in each view. Write the numbers on the dashboard.