Publish a portfolio dashboard without exposing source data

A portfolio should demonstrate modelling and analysis without releasing data you are not authorized to share. Use an original synthetic dataset or a dataset whose permitted use you have checked, and inspect the complete deliverable rather than only the visible report page.

Power BI's public-publishing option and authenticated sharing serve different audiences. Choose the distribution method deliberately before creating links or distributing a model file.

Build the demonstration on suitable data

The retail portfolio lab contains original synthetic customers, products and transactions, with reusable code and documented attribution terms. Its numbers are teaching examples, not real customer activity or NeuraPath business results.

This makes it possible to share the dataset and reproduce expected totals without asking readers to trust a private source. Label the synthetic nature prominently; do not write a case study suggesting that the dashboard improved a real company's revenue.

For a public third-party dataset, retain its source and applicable reuse terms. Publicly viewable does not automatically mean unrestricted redistribution.

Understand what public publishing exposes

Microsoft states that Publish to web makes reports publicly accessible without authentication and can expose underlying model data, including data not displayed on the visible page. It is unsuitable for confidential or proprietary data and is not a mechanism for relying on RLS to protect a public report. Publish to web documentation.

Hiding a column, hiding a page or removing a visual is not a reliable way to sanitize a model. Remove unsuitable data from the actual demonstration artifact before considering public distribution.

This article provides a preparation workflow only. The supplied lab does not publish a report, create an embed code or modify a service workspace.

Inspect every part of the artifact

Review imported tables, hidden fields, calculated tables, query steps, parameters, source paths, drill-through pages and tooltips. Remove credentials, tokens and internal connection details from anything distributed as source material.

A PBIX can carry more than a screenshot suggests. If you intend to share a downloadable project file, inspect its included data and metadata separately from the public report view. Do not assume that a private source connection makes the distributed model harmless.

Use a dedicated portfolio copy built from the approved demonstration source. Avoid starting with a production model and relying on a few visual deletions to remove sensitive material.

Choose a suitable review format

For an initial review, a local project package, screenshots or a recorded walkthrough may be sufficient. For interactive authenticated review, use the appropriate controlled sharing method and verify actual recipient access. Microsoft describes distribution choices in its sharing guidance.

For a genuinely public synthetic demonstration, confirm the tenant's allowed publishing options and review the final artifact before release. The fact that a feature is available does not decide whether the dataset is appropriate for it.

Include evidence of analytical quality

Package the problem statement, data dictionary, model diagram, measure definitions, source controls and acceptance results. For the retail fixture, demonstrate Paid value 69,500 paise, January 47,500 and the Unknown customer contribution of 5,000.

Show a controlled failure case, such as a duplicate dimension key or a drill-through population mismatch, and explain how the project detects it. This demonstrates practical reasoning more clearly than a gallery of unexplained charts.

Keep claims proportional to the evidence. A tiny synthetic model can demonstrate correctness and workflow, but it cannot establish production scalability, real-user adoption or business impact.

Verify the distributed experience

Inspect the exact file or link a reviewer will receive. Test navigation, filters, accessible labels and any allowed export paths. Confirm that no unpublished draft page or obsolete measure changes the intended story.

If a link is meant to require authentication, test it with the intended consumer context rather than only the author's signed-in session. If a file is meant to contain only synthetic data, inspect all included tables and not just the report visuals.

Exercise: create a portfolio release inventory listing every shared artifact and its data source. For each, state whether it is public, authenticated or local-only, and identify the evidence that its full contents are suitable for that audience.

NeuraPath's Data Analytics with Generative AI course connects portfolio projects with reproducibility and responsible delivery. A strong public example makes its data provenance, technical checks and limitations easy for a reviewer to verify.

Continue learning

This article is part of the Power BI data models and reporting sequence. Use the neighbouring tasks when you need the prerequisite or the next application.

• Review the preceding task in Design a stakeholder dashboard acceptance checklist.
• Return to the cluster foundation in Power BI star schema: define facts and dimensions first.