Data readiness means knowing where data lives, who owns it, whether use is authorized, how it is structured, updated and controlled. It need not be perfect, but limitations must be known.
Preparing data for an AI project: practical checklist
Access, quality, rights, sensitivity, freshness and document ownership. Reliable results require connecting technology to a workflow, data, an owner and a measure. The following principles structure that decision.
01 — Inventory sources, formats, volume, frequency and owners
Inventory sources, formats, volume, frequency and owners.
02 — Check usage rights, consent, contracts and retention periods
Check usage rights, consent, contracts and retention periods.
03 — Measure missing fields, duplicates, inconsistencies and time lag
Measure missing fields, duplicates, inconsistencies and time lag.
04 — Define a reference truth and correction process
Define a reference truth and correction process.
05 — Limit the prototype to minimum necessary data instead of copying everything
Limit the prototype to minimum necessary data instead of copying everything.
Action plan
Use this sequence as a starting point. Each step should produce a decision or verifiable output before the next.
- Create inventory
- Assign owners
- Classify sensitivity
- Test quality
- Define reference
- Document limits
Mistakes to avoid
- Waiting for perfect data
- Copying more than needed
- Ignoring date and version
Frequently asked questions
Yes for some uses, when format, quality and verification are understood.
A business owner for meaning and quality, supported by technical owners for access and security.
Key takeaway
Data readiness means knowing where data lives, who owns it, whether use is authorized, how it is structured, updated and controlled. It need not be perfect, but limitations must be known.
The important point is to progress through evidence: a precise use case, representative test, documented limits and an outcome-based decision.