Power BI can present operational results, but a report is not automatically a controlled data-entry workflow. When departments must submit monthly values, correct them, and prove who approved them, give that process its own application and SQL tables.
PHPRunner can build the submission and review pages over SQL Server. Power BI continues to read an approved reporting dataset. The two tools share governed data, but they have different jobs.

Quick answer: Write monthly submissions into a dedicated SQL table through a permission-controlled web application. Publish only approved rows to the Power BI reporting dataset, and treat “saved successfully” and “visible in the report” as separate confirmations.
Do not let operational users edit reporting tables or production facts directly. A submission table can store department, reporting period, metric or line, value, workflow status, revision, and the actors and timestamps for creation and change. A unique key should represent the one intended submission line for a department and period.
This design makes duplicates detectable and lets corrections remain visible. It also gives validation and approval rules a clear place to operate before data enters the reporting view.
A user's department access must be checked when rows are listed, opened, saved, exported, and submitted. Hiding another department in a dropdown is not security. Reviewers may need a broader scope, but it should be granted explicitly.
Use open and closed period states. Closing a month prevents ordinary edits; reopening it records who authorized the change and why. When two people edit the same submission, compare its revision or last-updated value and reject a stale save instead of silently losing the first change.
A useful workflow might move through draft, submitted, returned, and approved states. Power BI should read the approved reporting dataset, not every draft row. The application can confirm that a submission was saved or approved, but it must not claim that a report has already refreshed.
With Import mode, Power BI copies source data into its semantic model and needs a successful refresh to see later SQL changes. DirectQuery queries the source rather than importing the data, but caching, report behavior, permissions, and gateway configuration still need verification in the actual environment.
Decide whether a correction creates a new revision or changes an editable draft with a detailed audit record. Once a value has been approved and reported, a later correction should keep the earlier value, reason, actor, and time available for review.
Keep validation messages close to the business meaning: missing line, value outside an allowed range, duplicate department and period, or a closed period. Do not reduce the workflow to a generic form that accepts any number.
PHPRunner can create SQL Server data-entry, review, search, and administration pages with authentication and role-based permissions. Custom events and database rules can enforce period, department, approval, concurrency, and audit requirements.
This is a companion submission application, not native Power BI write-back. It also does not guarantee immediate report refresh or licensing savings. Confirm Power BI licensing, gateways, refresh limits, identity, and deployment architecture for the actual organization.