Original practice data
Messy contact list exercise
All names, addresses, and US 555 numbers are fictional. Work on a copy, then compare with the documented answer.
Data cleaning is not the same as making every cell look complete. A safe cleanup keeps uncertain values visible, applies documented standards, and never guesses personal information.
A safe method
Use seven steps in this order
- 01
Protect the source
Duplicate the file and keep an untouched original. Record the date and purpose of your working copy.
- 02
Profile before changing
Scan each column for blanks, inconsistent case, unusual formats, and possible duplicates.
- 03
Write the standards
Choose one display rule for names, phone numbers, and status values before applying edits.
- 04
Remove accidental spaces
Trim leading and trailing whitespace. Do not delete meaningful spaces inside names.
- 05
Normalize predictable text
Apply consistent capitalization and approved status values. Keep the raw copy for comparison.
- 06
Validate and deduplicate
Use more than a matching name: compare email, phone, and context. Flag uncertain records for review.
- 07
Document every exception
Note missing values and unresolved errors instead of silently inventing replacements.
Visible transformation
Before and after
We changed Amira's status to “Needs review” because the correct domain is unknown. Guessing would create false data.
Data-quality exercise
Can you find the six issue types?
Download the practice file and inspect it before revealing the review guide.
Explain your decisions
Make the cleanup auditable
Keep the untouched source, your cleaned copy, and a short change log together. The change log should name the columns reviewed, the standards you applied, the number of values changed, and every unresolved exception. For a portfolio sample, show three representative transformations rather than publishing a large contact table. Explain why trimming an accidental space is safe, why changing a known status to an approved label is controlled, and why an incomplete email must remain flagged instead of receiving a guessed domain.
A useful final check is to sort and filter the cleaned data, compare the row count with the source, and confirm that no record disappeared unintentionally. Record the application used, the date of the test, and whether formulas were converted to values. This evidence demonstrates data judgment and traceability. It does not imply that you handled real customer or employee information.
Sources and review notes
This guide demonstrates data-quality decisions rather than a changing software interface. Formula names and menu locations should be checked against the application version used by the learner.
- Practice and answer files: original First Role Skills examples
- Byline: First Role Skills Editorial Desk
- Reviewed: 3 August 2026
- AI assistance: structure and drafting; editorial accountability remains with the publisher