Practical framework / 5 MIN READ
Make the report say how current its evidence is
Separate a dashboard refresh from the age and completeness of its source data.
AI-assisted research and analysis for the SNS Creative Fieldbook
Sources checked 2026-11-06
A report can refresh successfully while displaying yesterday’s upstream data. Useful freshness expectations connect source delivery, transformation completion and the decision’s deadline rather than relying on a single green status icon.
What you will leave with: Create a freshness agreement with source-specific checks, visible status and a fallback decision rule.
Use this when
- A dashboard says updated now but recent orders are missing.
- Teams disagree about whether morning reporting is ready for use.
This lesson defines reporting service expectations, not a guarantee that recently loaded data is accurate or complete.
Define ready for the decision
Identify who uses the report, when they need it and what lateness would change. A weekly planning review may tolerate a longer delay than an operational stock check. Avoid promising real-time reporting where the underlying provider updates only periodically.
For each source, state the expected delivery schedule, timezone and known exceptions. Distinguish the latest event time, latest successful load and latest completed report build. A quiet source may have no recent events despite a healthy pipeline, while one recent event may arrive even when most of a batch is missing.
Measure freshness at the right stage
dbt documents source freshness checks using loading information or warehouse metadata and configurable warning or error conditions. Such checks can reveal stale inputs, but they do not alone establish that every expected record arrived. Define the field’s meaning before using it as a clock.
Pair source freshness with delivery-status or expected-partition checks where appropriate. Confirm the transformation and dashboard actually consumed the fresh input. A successful connector run and a successful visual refresh are separate events; neither should silently stand in for the entire chain.
Show status where decisions happen
Display the evidence cutoff and status beside decision-critical results, including exports. Use plain language such as provisional through yesterday evening or delayed order feed, rather than a generic updated label. Explain which metrics are affected and which remain usable.
Choose what happens when the threshold is breached: show a warning, withhold a comparison or use the last reconciled period. Do not silently mix fresh spend with stale revenue and present the resulting efficiency ratio as current. Preserve a consistent fallback view so readers do not improvise incompatible workarounds.
Assign ownership and rehearse a delay
Name the source owner, report owner and person who decides whether the report can support the meeting. Set a route for material delays and a review time. Avoid notifying every stakeholder about every routine retry; the message should explain the impact on a decision.
Test the policy with a simulated delayed feed and a partial batch. Check that the warning appears, the fallback works and users know what to do. Review actual missed expectations periodically and adjust either the pipeline or the promise, rather than normalizing an inaccurate green status.
Make the reasoning concrete.
A hypothetical dashboard refreshes at 09:00, but the order feed is complete only through 18:00 the previous day while media spend is current through midnight. The displayed acquisition ratio mixes unequal periods.
Trust the refresh badge
Opening: Use the current ratio in the morning budget review.
Hypothesis: The dashboard job succeeded.
Risk: Revenue and spend cover different evidence windows.
Use an aligned fallback
Opening: Show the last jointly complete period with a visible feed-delay note.
Hypothesis: The decision uses compatible data.
Risk: The newest hours remain unavailable until recovery.
Use the aligned period and identify the delayed source. Revisit the decision when the feed is reconciled if the missing interval could materially change it.
Take the questions into your next brief.
Use these prompts in your own notes. Nothing is submitted from this page.
Decision time
When must which evidence be ready?
Clocks
What do source, load and build timestamps mean?
Completeness
What checks detect partial delivery?
Fallback
What view remains safe when the promise is missed?
Owner
Who resolves the delay and communicates its impact?
Watch for the wrong conclusion.
- Equating visual refresh with fresh source data.
- Using a single recent record to prove complete delivery.
- Mixing differently aged numerator and denominator.
- Leaving a stale dashboard marked healthy.
Your next action
Define a visible evidence cutoff and a tested fallback for one report used in a time-sensitive decision.
Sources & scope
Frameworks, worksheets and hypothetical examples are original editorial synthesis. Linked primary sources support only the stated concepts. Source checking is not an independent audit of effectiveness. No named human author or reviewer is claimed.
- dbt: Add sources to your DAG, source data freshness →Primary technical documentation · Supports: Documents source freshness checks using loading information or metadata with configured warning/error conditions. Freshness alone does not prove completeness or correctness.