Help & user guide

Getting started

Tabular Trace maps the dependencies inside your Power BI estate — from the original source, through dataflows and the semantic model, out to every report visual — so you can see what a change will break before you make it.

Your first five minutes

• Create a project from the Projects page. Give it a name and, optionally, a company name — a team can manage projects for different companies.
• Upload a file: a Power BI project folder export (.pbit or PBIP files), a dataflow export, or a SQL Server / SSIS / dbt / Fabric artifact.
• Open the project and explore the lineage views. Every node and edge was parsed from your files — nothing is invented.
• Try the Preview tab to see your report pages reproduced from saved layout metadata, and the Reports tab for health, inventory, and impact summaries.

Parsing happens securely on the server and never throws on a bad file — a file that cannot be fully parsed is recorded with a warning instead of failing the whole upload.

Uploads & supported technologies

Tabular Trace traces these technologies:

• Power BI reports — PBIP folders and .pbit files (report pages, visuals, filters, bookmarks).
• Semantic models — TMDL or .bim files (tables, columns, measures, DAX, relationships).
• Dataflows — Power Query M is fully parsed, step by step.
• SQL Server — scripts and Agent jobs.
• SSIS — package archives.
• dbt — sources and models, stitched to your SQL Server and semantic-model lineage.
• Microsoft Fabric — pipelines, lakehouses, warehouses, notebooks, and OneLake shortcuts.

Tips

• Upload the model and the report as separate files — report uploads contribute pages and visuals; model uploads contribute tables, columns, and measures.
• Use the “Model only” switch on a report upload when you want its model content without its report pages.
• Uploading the same file again replaces the previous version — the header shows how many files were replaced. Old versions rest in the recycle bin.
• Friendly names: every upload keeps its original file name, and you can give it a friendlier display name on the Files tab.

Lineage views

Every project offers five views of the same lineage graph. Arrows always point downstream: source → dataflow → model → visual.

• Lineage graph — the full map. Click any node to inspect it; click an edge to trace the path it belongs to.
• Field lineage — follow a single column or measure end to end, with a chart layout that minimizes crossing lines.
• Visual trace — start from a report visual and walk back to its sources.
• Names — every ID found in your files, resolved to a friendly name wherever the artifacts themselves provide one.
• Gaps — anything the lineage could not resolve, grouped by what to upload next, with severity levels and guidance.

Search highlights matches without disturbing the layout. Filters only light up as active when they actually restrict the data shown.

Report preview

The Preview tab reproduces your report pages from the saved layout metadata: positions, sizes, z-order, visual types, fields, and formatting. It never touches your source data — every value on the canvas is a deterministic synthetic placeholder, which is what the “Placeholder data” badge reminds you of.

• Pages hidden in Power BI are labelled “· hidden” in the page selector but stay selectable.
• Hidden visuals can be revealed with the “Show hidden visuals” toggle.
• From any report page in the lineage views, use “Open report preview” to jump straight to that page.
• Files uploaded before hidden-page detection need a re-analysis (Files tab) to gain the hidden labels.

Reports & exports

The Reports tab turns the lineage graph into documents you can share:

• Health, inventory, impact, and history — project-level reports with Direct/Indirect tags and a “What to do” box.
• Dataflow reports — per-dataflow diagrams from source through each Power Query step to the published entity, with numbered M steps and copyable code.
• Build guides — searchable, linked, collapsible HTML with a light/dark toggle, alongside a Word export.
• All project reports — the team-level view across every project, with Excel export.

Most reports export to HTML and Word; the team report also exports to Excel.

AI Q&A

“Ask the report” answers questions about your lineage in plain language — what feeds a visual, what a change to a column would hit, where a gap comes from. Answers are grounded in your parsed lineage, not general knowledge.

• AI questions are metered: the Pro plan includes 100 questions per month.
• Identical questions are answered from a cache, so repeats don’t consume your allowance.
• Everything else in the app — parsing, lineage, previews, reports — is deterministic and uses no AI credits.

Teams, projects & housekeeping

• A team owns projects; invite teammates from the Team page. Company name is stored per project, so one team can manage projects for different companies.
• The recycle bin keeps soft-deleted files and projects — restore them or purge them permanently. Team storage totals are shown under “Team storage”.
• The activity log (More → Activity log) records what changed and when.
• Power BI tenant sync (More → Power BI tenant sync) discovers artifacts straight from your tenant when you have the required tenant settings enabled.

Impact search

Impact search answers “if I change this, what breaks?” across every project in your team. Search for a table, column, or measure and see every downstream dependency, tagged Direct or Indirect, with the path that connects them.

For BI developers

A few notes for technical users:

• Power Query M is parsed step by step — every step is numbered and its code is copyable from dataflow reports.
• DAX expressions are parsed for column and measure dependencies, including calculation groups.
• dbt projects merge into the same graph: sources and models are stitched to SQL Server objects and semantic-model tables once both sides exist.
• Fabric notebooks contribute inferred edges; pipelines, lakehouses, warehouses, and shortcuts are first-class nodes.
• GUIDs from tenant exports are resolved to names using the metadata in your own uploads — the registry never invents names.

FAQ

Is my data safe?

Yes. Files are parsed on the server, secrets are redacted before storage, and report previews use synthetic placeholder values only — your source data never appears in a preview.

I uploaded the same file twice — where did the first one go?

The new upload replaces the old one (the header shows “x replaced”), and the replaced version is in the recycle bin if you need it back.

Why is a page or visual missing from the preview?

Model-only uploads deliberately exclude report content. If a report was uploaded with “Model only” switched on, its pages won’t appear — re-upload with the switch off.

What does a lineage “gap” mean?

A gap is a dependency your files reference but don’t define — usually an artifact that hasn’t been uploaded yet. The Gaps view groups them by what to upload next.