Semantic Model Readiness Checklist
10 questions to answer before you build, or rebuild, your reporting
With an AI partner, building reports is the easy part. Whether those reports can be trusted depends on the semantic model underneath, and the model depends on how well the business is understood. Answer yes or no to each question.
Business understanding
- 1.
Can you name the three to five decisions this reporting must support, and who makes them?
Reports built without a decision in mind become reports no one opens.
- 2.
Do your key metrics each have one written, agreed definition?
If "revenue" or "on-time" means different things to different teams, every dashboard starts an argument.
- 3.
Do you know which business processes produce the data?
To improve any process, you must be able to measure it, and you can only measure what you understand.
Data
- 4.
Do you know the system of record for each core entity (customers, products, employees, projects)?
Two sources of truth means no source of truth.
- 5.
Can you get the data at the level of detail you need (each order line, each day)?
A model can roll detail up; it can't invent detail that isn't there.
- 6.
When the same number appears in two places today, does it match?
If it doesn't, and no one knows why, unexplained differences are the fastest way to lose trust in BI.
Model
- 7.
Are your reports built on one shared model, rather than each carrying its own copy of the data and logic?
Shared models mean one fix, everywhere.
- 8.
Is your business logic written once in the model, not in spreadsheets or report visuals?
Logic scattered across reports can't be reviewed, tested or reused.
Ownership
- 9.
Could someone on your team explain and change a measure without the person who built it?
If not, you don't own your BI; you rent it from one person.
- 10.
Is the model version-controlled and documented, so changes can be reviewed?
Undocumented models get rebuilt instead of improved.
Count your yes answers
- 8–10
- You're ready to build. The model can come together quickly.
- 5–7
- Shore up definitions and sources first; it will save rework.
- 0–4
- Start with discovery. A Power BI Health Check is a good first step.
Get the PDF
A printable copy with yes/no boxes, sent to your inbox.
Want a second opinion on your answers? Start a conversation.