Adovo

Data & Analytics

The Power BI model refreshed, but the subscriber got no report

A successful semantic model refresh does not prove that the after-refresh email reached anyone. Separate refresh and delivery evidence.

October 26, 2026

The dataset owner sees a completed semantic model refresh. The executive still did not get the morning report. A Microsoft Fabric Community issue reports subscriptions that were delivered only after retries despite completed refreshes. The report is an observation from a community member, not a confirmed platform-wide defect. It illustrates a broader operating gap: refresh and delivery are different events.

Prove each stage separately

Record the refresh ID and completion time, the subscription's configured frequency, the expected recipient list and the actual delivery evidence. Microsoft explains that an After Data Refresh subscription sends after the first scheduled refresh of the day, which may differ from a manual or later refresh. Check whether the report uses a live connection, whether the subscription is active, and whether the recipient or tenant policy blocks delivery. A report-page refresh does not refresh the underlying model.

Test the handoff

Run the subscription's own test action if available and compare the result with its after-refresh behavior. Note if delivery is delayed or absent; Microsoft says peak demand can delay queued mail. If retries appear necessary, preserve timestamps and raise a support case rather than designing an undocumented retry loop around a presumed product bug.

Make the business deadline visible

For a report that drives a daily decision, set a deadline for receipt, not only for data freshness. Maintain a simple check of expected versus received delivery and an owner for escalation. If the email is not the reliable control plane, offer a stable report link and a clear freshness timestamp so the decision maker can still inspect the data.

Adovo’s view

A green refresh says the model was processed. It does not say the decision reached its owner. Measure the last mile at the point where someone uses the information.

Continue exploring

Data & Analytics

Microsoft FabricCapacity

Fabric capacity alerts arrive after users complain: monitor the risk window

Capacity utilization alone is a weak incident trigger. Combine near-current health, throttling stages and item attribution into an operations runbook.

Read more

Data & Analytics

Microsoft FabricDirect Lake

A Direct Lake model moves workspaces but still points at the old source

A copied or Git-deployed semantic model can retain a source binding. Verify the lakehouse endpoint and identity before rollout.

Read more

Data & Analytics

Power BISemantic models

How to prove a Power BI report update did not change the numbers

Screenshot comparison misses semantic regressions. Bind each Power BI release to a source snapshot, identity and filter matrix, and expected measures.

Read more
Explore data and analytics services →