Quality
Data freshness
Also called recency.
Data freshness is how recently the numbers in a report were updated. A report can be right and still be old.
How it is measured
Compare the newest hit in the report with the current time. The gap is the freshness. Reports built from a nightly batch are a day stale, live views are seconds stale.
Tarsier's Live view reads recent hits within seconds. Historical reports rely on aggregation and today's numbers can still move as late pings arrive.
Worked example
A sale launches at 09:00. By 09:20 the live panel shows 61 active users, but the day's chart reads 14 visits. The owner thinks the sale failed. The aggregated chart is eight minutes behind and the 14 reflects only the first chunk.
At 10:00 the two agree within a few percent.
How it differs
Data freshness is the state of the report. Lag is the delay that causes it. Processing latency is the time inside the pipeline. Real-time analytics is the tuned-for-freshness style.
Common errors
Judging a launch from a stale chart. Comparing a live view with a daily one. Forgetting time zone of the cut-off. Assuming yesterday is final. Hiding the update time.
In practice
Show the last update time on every dashboard and check it before reacting to a sudden dip.