Fill in these first
- Dataset name and purpose: [e.g. orders sheet used for weekly sales reports]
- Column headers: [paste the header row]
- Sample rows: [paste 10 to 20 rows with personal details removed]
- What you know about the columns: [notes or none]
- Who maintains the data: [team or role]
Task
Write a data dictionary that a new person could use to read and query this data correctly.
Method
- Infer the type and format of each column from the sample rows.
- Where the meaning is not obvious, give the most likely meaning and mark it "to confirm".
- Look for mixed formats, blank values, codes without a key and columns that seem to duplicate each other.
Output format
| Column | Plain meaning | Data type | Format or unit | Allowed values or range | Example | Can be blank | Status (confirmed or to confirm) |
Then:
- Data quality questions to ask the owner, one per unclear column.
- Suggested column renames for any confusing names.
Rules
- Never copy personal data from the sample rows into the examples; use made-up values in the same format.
- Dates state their format, such as DD-MM-YYYY.
- Money columns state the currency and whether tax is included, or mark it "to confirm".
- Do not guess meanings for codes; list them for the owner.
Example
order_status | Current stage of the order | Text | Code | NEW, PAID, SHIPPED, RETURNED | SHIPPED | No | To confirm






