Check if your reporting metadata
is ready for AI
Before your data can answer questions reliably, it needs clear names, plain-English definitions and consistent structure. Starting with Power BI-style metadata, this scanner checks those foundations and produces a readiness score, suggested improvements and a practical action plan.
No sign-up required · Results in under 30 seconds
The problem
AI tools are only as good as your metadata
Natural language querying, Copilot-style experiences and self-service analytics all depend on the underlying field names, descriptions and structure being understandable — not just to developers, but to the people and tools asking questions.
What goes wrong
- Cryptic field names like FACT_cust_rev_MTD
- Measures named GP%, OrdersCnt, Adj_EBITDA
- Fields called Amount, Value or Total with no context
- Missing descriptions on every field
- No synonyms for natural language querying
- Inconsistent casing and abbreviations across tables
What happens as a result
- AI tools guess the wrong meaning and return incorrect answers
- Business users cannot find the field they need
- Copilot and Q&A return "I don't understand" responses
- Self-service reporting stalls and developers get repeat queries
- Data teams spend time explaining what fields mean instead of building
- Ambiguous measures cause debates after the report is live
What good looks like
- Clear, plain-English field names with no abbreviations
- Every measure named for what it means, not how it's calculated
- Descriptions that explain each field to a business user
- Synonyms that map natural language phrases to the right field
- Consistent naming conventions across all tables
- AI tools that return reliable, accurate answers to real questions
What gets checked
Eight dimensions of readiness
The scanner evaluates every field across these dimensions and combines them into an overall score.
Field and column names
Flags cryptic names, database prefixes (FACT_, DIM_), abbreviations, underscores and names that would confuse a business user or AI tool.
Measure names
Checks whether measures use clear business language — not internal shorthand like rev_mtd or GP% that AI tools cannot reliably interpret.
Descriptions
Checks which fields have descriptions and which are blank. Missing descriptions prevent AI tools from understanding what a field actually means.
Synonyms
Identifies fields that lack synonym coverage — alternative words a user might type when asking a natural language question.
Ambiguity and vagueness
Flags fields that could mean multiple things (Amount, Value, Total) and marks them for business confirmation before publishing.
Abbreviations
Catches shortened names that users and AI tools will not understand — Qty, Cnt, Rev, Pct, MTD, YTD and similar patterns.
Naming consistency
Identifies mixed naming conventions across the model — some fields using underscores, some PascalCase, some abbreviations — which reduces AI reliability.
AI and Copilot readiness risks
Combines all signals into a risk assessment highlighting the specific issues most likely to cause failures in AI-assisted querying or natural language reporting.
How to provide your metadata
Four ways to scan
The scanner works with Power BI metadata in the formats you already use — no special export required.
Paste field names
Copy field names directly from your Power BI field list and paste them in as plain text. The simplest way to get a quick readiness check.
Upload a CSV export
Upload a metadata export from Tabular Editor, DAX Studio or any tool that produces a CSV of fields, measures and descriptions. The scanner auto-detects the format.
Paste TMDL or BIM JSON
Paste TMDL (Tabular Model Definition Language) or BIM JSON directly. This gives the deepest analysis — including expressions, display folders, format strings and descriptions.
Upload a screenshot
Upload a screenshot of a dashboard, report or field list. Claude Vision extracts the visible field names and passes them through the scanner.
Scan output
Seven result sections. Three export formats.
Every scan returns a structured set of outputs you can act on, share with your team, or hand to a developer.
Score
- Overall AI readiness score from 0 to 100
- Score breakdown across four categories: Field Naming, Business Language, AI Compatibility and Consistency
- Ambiguity risk level — Low, Medium or High
- Plain-English summary of strengths and what will confuse AI tools
Renames
- Suggested new name for every problematic field
- Plain-English reason for each suggestion
- Confidence level: High, Medium or Low
- Confirmation flag for ambiguous fields that need business input
Descriptions
- Tooltip-ready description for every field, written for your audience and report area
- Confidence level on each description
- Confirmation questions for fields where the business definition is unclear
Synonyms
- Alternative words and phrases a user might type when querying this field
- Improves phrase matching in Copilot, Q&A and other AI-assisted querying tools
Display folders
- Suggested report groupings for organising fields in the field list
- Helps users find the right field quickly and reduces AI guessing
Business questions
- Example questions a business user or AI tool should be able to answer after the renames are applied
- Useful for UAT — paste into Copilot or Q&A to validate the improvements
Action plan
- Prioritised checklist of exactly what to fix — sorted High, Medium and Low priority
- Focused on the changes that will most improve the readiness score
Export formats
- Markdown report — copy to clipboard or download for sharing
- Naming fixes CSV — current name, suggested name, reason, confidence
- Descriptions and synonyms CSV — ready to import or paste into documentation
When to use it
Common use cases
Preparing a model for AI-assisted querying
Before enabling Copilot, Q&A or any natural language feature, scan the model to identify which fields will cause failures or incorrect answers.
Reviewing an inherited report
When you take ownership of someone else's Power BI model, run a scan to understand the current naming quality before making changes.
Cleaning up before publishing
Use the scan results to create a backlog of naming fixes, descriptions and synonyms to apply before the model goes live.
Improving self-service reporting
Clear field names and descriptions help business users find what they need without developer support — the scan shows exactly what is currently blocking this.
Preparing for a manual audit
Run the free scan first, then share the results with a reviewer. The scan output becomes the starting point for a deeper manual review.
Creating a naming standards baseline
Use the score and recommendations to set a minimum readiness standard for new reports and models across your team.
Checking a model before a data product launch
Before making a semantic layer or data product available to a wider audience, confirm it meets readiness standards for AI and self-service use.
Part of a wider journey
Step one in the ReadyMyData workflow
The AI Readiness Scanner covers the Assess stage. The wider ReadyMyData platform is building tools to support every stage of a data project.
Assess
This is where the AI Readiness Scanner sits. Understand your current readiness before planning any changes.
Define
Use the Requirements Gathering tool to capture KPIs, audiences and report structure with stakeholders.
Design
Generate a semantic model blueprint and report structure from your confirmed requirements.
Deliver
Create tasks, UAT checklists and sign-off packs to complete delivery correctly.
Improve
Re-scan after changes to confirm readiness improved and track progress over time.
A transparent note on current scope
The AI Readiness Scanner currently works best with Power BI-style metadata — field names, measures, TMDL exports, BIM JSON, CSV metadata files and dashboard screenshots. The analysis focuses on naming quality, description coverage and semantic structure for AI-assisted querying in Power BI-style environments.
ReadyMyData is designed to expand into a broader data project delivery workspace over time — supporting more metadata formats, data tools and delivery workflows. The scanner is the first live tool in that journey.
Ready to check your metadata?
Free, instant, no sign-up. Paste your field names or upload a metadata export and get your results in under 30 seconds.