Skip to content
FREE TEMPLATE

AI data readiness worksheet

Check sources, quality and access for one specific use case.

What you’ll prepare

Sources, defects, remediation actions and an approved sample.

When to use it

Before estimating or starting an AI trial.

Fill in the fields below and download your document. The website does not send or save answers: closing the page may lose them. Examples are illustrative and should be replaced with your process data.

Download the complete blank template (.md)

1. Identify the sources

Start with information needed for the intended decision or action. Exclude irrelevant archives and record who understands the meaning of each source.

See a worked example

Illustrative example: technical procedures in a shared folder, reviewed quarterly by the support owner. Personal copies and unapproved drafts remain outside the trial.

Record the system, operational owner, available format and stable record identifier.

Describe who may access the data, for which activity and who authorises its use in this project.

Record update frequency, obsolete information and internal removal rules that need confirmation.

2. Inspect a sample

Select common and difficult cases. Record the selection method: a sample containing only clean files hides the conditions encountered in actual work.

See a worked example

Illustrative example: 40 documents cover digital PDFs, scans and tables. Five have conflicting versions: identify the authoritative version before constructing the evaluation set.

Describe period, size and variety: languages, formats, teams, rare cases and incomplete documents.

Count missing fields, duplicates, conflicting versions and records that cannot be matched.

Identify who can explain fields and establish the expected answer when information is ambiguous.

3. Plan the preparation

Separate issues that prevent a trial from manageable limitations. Assign concrete actions without waiting to clean every record in the business.

See a worked example

Illustrative example: begin with one approved document family. Unreadable scans go to manual processing and are counted separately when results are reported.

For each gap state what changes, who acts and how completion will be checked.

Define categories, permissions and cases that must stay outside the initial trial.

Specify which sources, labelled examples and authorisations must exist before the trial starts.

Pitfalls to avoid

Check before using it
  • Confusing technical access to a file with permission to use it.
  • Assessing quality only on easy cases.
  • Launching an assistant with conflicting versions and no authoritative source.

Let’s discuss your project.

Review the text before sending. Only your email is needed alongside the message; other details are optional.