Why this matters
Real data is messy. Field forms have duplicates, registers have spelling differences, and exported files mix numbers with text. Several adverts list data cleaning and validation as core duties, and analysts often say cleaning takes more time than analysis itself. If your data is wrong, every chart and report built on it is wrong too.
Work on a copy
Never clean the original file. Make a copy, name it clearly (for example survey_clean_v1) and keep the raw download untouched. If you make a mistake, you can start again.
Common problems and fixes
Duplicates. The same household submitted twice, often because a phone uploaded twice. Look for repeated IDs or identical rows. In Google Sheets, Data cleanup then Remove duplicates helps, but check what it removes first.
Inconsistent spellings. "Tamale", "tamale", "Tamale Metro" and "TML" may all mean the same place. Use TRIM to remove extra spaces and PROPER or LOWER to fix capitals, then agree one standard name and replace the others with Find and replace. A filter on the column shows every distinct spelling at once.
Missing values. A blank can mean "not asked", "refused" or "forgot to enter". Find out which. Do not fill blanks with zero or a guess. Record how many are missing and decide with your supervisor whether to follow up, leave them blank or exclude that record from a particular calculation.
Impossible or unlikely values. A child aged 45, a farm of 9,000 acres, a date in the future. Sort each number column from smallest to largest to spot extremes quickly. Some extremes are real, so check with the field team before changing anything.