Setting data freshness for reports: Define required source cutoff for decision-making events.; Set ready-by time and visible status for late or incomplete periods.; Track source event time, integration receipt, dataset inclusion and report visibility.
Image: Business Insight Stack

Metric Definitions

Part of Data integration for BI

Defining data freshness expectations for a report

Specify a BI report's source cutoff, ready-by time, completeness rule and reader-facing status when data is late.

Desk Governance · · 3 min read

Define freshness by the source events a report must include when someone uses it. State the required cutoff, when the result should be ready and what readers will see if it is late. A recent refresh timestamp alone does not establish that the source period is complete.

Write the expectation

Ask the report owner when the figure informs a decision and how late a change can arrive before it matters. Record the event time in the source, when integration received the event, when the reporting dataset incorporated it and when a reader could see it. These points answer different questions.

For a Monday backlog review, an owner might require eligible case changes through a defined Sunday cutoff before the meeting. The team must also agree on the reporting timezone, reopened cases and how late-entered changes revise an earlier period. That cutoff is illustrative, not a general standard.

Agreement field / Question it answers

Required source cutoff
Which events must be included?
Ready-by time
When does the reader need the result?
Complete-source rule
Which deliveries must arrive before release?
Visible status
How are late, partial and unavailable periods shown?
Owner and response
Who investigates a breach and decides what to display?

Measure the visible result

A successful integration run can copy no new data because nothing changed, or because an expected delivery was absent. A recent source event can reach a destination while a report still shows an older model.

In Power BI Import mode, source changes require a semantic-model data refresh before they appear in the imported copy. Keep the model update time alongside the source coverage it contains.

For a streaming path, distinguish transport metrics from report freshness. Datastream's data-freshness metric measures delay until it reads a source event; total latency measures delay until that event reaches the destination.

The freshness metric excludes events not yet read and is set to zero when there are no new events to read. A zero reading therefore cannot establish that all expected events reached the report.

A source-freshness check can help when its load timestamp or query represents actual arrivals. dbt supports warning and error thresholds for configured source freshness. Pair such a check with the expected delivery schedule: a legitimate quiet period and a missed file need different interpretations.

Data Freshness Metrics: Source vs. Reporting Layer

Source-Freshness Metric (e.g., Google Datastream)
Delay from source event to ingestion
Total Latency (Streaming Path)
Delay from source event to destination delivery
Report Freshness (e.g., Power BI Import Mode)
Time of semantic model refresh after data arrival

Give readers a usable status

Near a consequential figure, show the business period covered, latest included source cutoff and whether the period is complete. If one feed is late, identify the affected measure or population. If the report retains the last accepted result, keep its earlier cutoff visible.

Agree in advance whether a breach holds the affected result, permits a clearly marked partial view or retains the last accepted one. Check the rule against a quiet source period, a missing delivery, a late correction, a failed transformation and a delayed BI refresh. The status shown to readers should distinguish those situations where they change the decision.

Handling Late or Missing Data in Reports

  • Pros of holding the affected resultMaintains consistency; avoids misleading partial views
  • Cons of holding the affected resultMay delay critical insights if data is significantly delayed
  • Pros of allowing a clearly marked partial viewProvides timely insight with transparency about limitations
  • Cons of allowing a clearly marked partial viewRisk of misinterpretation if readers don’t understand the caveat
  • Pros of retaining the last accepted resultEnsures continuity; reduces confusion during outages
  • Cons of retaining the last accepted resultMay not reflect current reality, especially in fast-moving environments

More from Metric Definitions