Define a Spreadsheet Cleanup Job Before Changing the Data

A spreadsheet cleanup service should start with a preserved source file and a written definition of what will change. Agree on the sheets, columns, duplicate rules, formats and checks before making edits. A tidy-looking table can still contain the wrong records, broken formulas or a change the owner did not authorize.
This guide proposes a small cleanup workflow using invented practice data. It is not a procedure for auditing financial accounts or repairing a business system whose dependencies you have not inspected.
Identify the workbook's purpose
Ask what someone needs to do with the file after delivery. A list used to contact event attendees differs from a history of every registration attempt. Repeated names may be an error in one file and useful records in the other. The purpose determines which questions need answering before cleanup.
Inventory the included sheets and the columns that identify a record. Ask about formulas, linked files, hidden sheets, protected areas and exports consumed elsewhere. Mark anything you cannot evaluate. A request to standardize a contact list is not permission to rewrite calculations or remove an unfamiliar worksheet.
Agree on the intended application and delivery format. A working file, an import file and a printed report may have different requirements. Keep access limited to the data needed for the job, and ask for an authorized sample when the full workbook contains information outside your scope.
Preserve a source and define the change rules
Keep the supplied file unchanged and work on a separate copy. Record its version and the date it was received. If the owner sends a replacement, compare the scope before mixing the new rows into work already reviewed.
Write a short rule list. It might specify consistent category labels, an agreed date display and a review group for possible duplicates. Include examples of acceptable and unacceptable changes. Do not silently correct a value because it looks unusual; the unusual value may represent a real exception.
Choose how you will record changes. For a small file, an issue list with the sheet, row identifier, proposed action and decision may be sufficient. Avoid depending only on row numbers when sorting could change them. Retain a stable way to connect the working record with its source.
Review duplicates before removing them
Microsoft's Excel guidance distinguishes highlighting duplicate values for review from using Remove Duplicates, which permanently deletes duplicate data. It recommends preserving a copy before removal. The choice of columns matters because the tool's comparison needs to match the records you intend to treat as duplicates.
Source: Microsoft's duplicate review and removal guidance. Read 2026-10-26; confirm the controls in the client's actual Excel version.
For an invented event list, two entries might have the same name but different booking references. That is not enough evidence to discard either booking. Conversely, a repeated reference might be a duplicate import or an amended registration that should remain traceable. Ask the owner which interpretation is correct.
Create a review group before proposing removal. State the columns compared and the reason the rows were flagged. Keep confirmed duplicates separate from unresolved matches. A tool can identify repeated values, but it does not decide the business meaning of a record.
Use validation as an entry rule
Microsoft describes data validation as a way to restrict the type or values entered into cells and provide input messages or error alerts. Its examples include lists, numbers and dates, and it recommends testing valid and invalid entries. That feature helps apply an agreed rule; it is not independent proof that the underlying data is true.
Source: Microsoft's data validation instructions. Read 2026-10-26.
Before adding a rule, agree on the allowed values and how genuine exceptions will be handled. A category list that excludes a valid case can encourage people to choose an inaccurate alternative. Include a clear route for asking about an unfamiliar entry instead of forcing every row into a convenient label.
Document where the rule applies. Test an ordinary valid value, an invalid value and an empty entry if blanks have a defined meaning. Check the actual behavior in the required application; do not promise that a rule will behave identically in every export or spreadsheet program.
Check old values and copied entries separately
Do not treat a newly added rule as a completed check of old values. Microsoft's advanced Excel guide says existing invalid entries do not automatically produce a notification. In the authorized practice copy, select the agreed range and use the desktop workflow it describes: Data > Data Validation > Circle Invalid Data. Review the flagged values against the owner's rules; a circle identifies a question to resolve, not permission to replace a value.
The same guide says validation messages do not appear when data is copied or filled. Include a small copy or fill example in the practice file, then check its resulting values separately against the rule. Record which entry routes were checked in the handover, so a successful typed-entry test is not mistaken for verification of a copied batch.
Source: Microsoft's advanced data validation guidance, checked November 15, 2026. Its listed versions are Excel for Microsoft 365, 2024, 2021, 2019 and 2016; confirm the commands in the recipient's application.
Protect identifier columns before import
Decide which columns contain identifiers before opening an incoming file. Microsoft's Excel guide explains that automatic conversion can remove leading zeros, and numeric values have a maximum precision of 15 significant digits. Codes with 16 or more digits need text storage to retain every digit. Changing the format after digits have been lost does not recover the original identifier.
For a fictional product list, 00123 and 123 might identify different items. If conversion leaves both as 123, a duplicate review can no longer distinguish them from that column alone. Return to the unchanged source and confirm the code rules with the owner. Adding zeros to match an assumed length can create a plausible code without proving it was the original.
Microsoft's CSV guidance says opening a CSV uses Excel's current default data formats to interpret its columns. Where the recipient's Excel supports it, use Data > From Text/CSV and inspect the import through Transform Data; set identifier columns to Text before loading. Compare example codes with the source, then reopen the agreed delivery format and check them again. Record the import method in the handover.
Sources: Microsoft's identifier and number-precision guidance and Microsoft's CSV import instructions, checked November 16, 2026. Commands vary by Excel version; this is a planning check, not a report of a tested client workbook.
Build a practice cleanup with an issue log
Create an original fictional list of workshop registrations. Include a repeated reference, inconsistent category spelling, a missing date and two legitimate bookings under one name. Keep the source copy and write the proposed rules before editing.
Work through the sample and record each decision. Standardizing an agreed spelling can be straightforward, while deciding which booking is current requires more information. Leave the unresolved case visible rather than inventing a date or reference to make the file appear complete.
Compare the result with your own rule list. You should be able to explain which records changed, which remained untouched and which still need review. The exercise is intended to make the proposed process reviewable.
Check records and dependent content
Compare source and delivered counts using the agreed scope. Explain differences rather than treating a smaller file as proof of successful cleanup. Confirm that retained records still contain the fields the owner needs, and review a sample of changed and unchanged rows.
If formulas or links are included in the assignment, check their expected behavior separately. A displayed value may look plausible even when a reference has changed. Use agreed examples with known outcomes and mark anything you cannot validate. Do not describe a formatting review as a complete calculation audit.
Open the saved delivery file and inspect it again. Check sheet names, column order, data types and any required import layout. A file that looked correct before saving or exporting may not be the same deliverable the client will open.
Check the CSV delivery scope
Agree which sheet a CSV delivery must contain. Microsoft's text-format guide states that Excel's listed CSV formats save only the active sheet; it also warns that saving in a text format removes all formatting. A CSV export from a workbook with several sheets therefore needs its own delivery check.
For the fictional workshop file, the registrations sheet and the issue log may be separate sheets. Keep the working workbook, then export the agreed sheet under a distinct filename. If the client also needs the issue log, agree how to deliver it separately. Open each promised file and compare its headers and record count with the intended sheet. These are proposed handover checks, not results from a tested client file.
State which file is the working workbook and which is an import copy. If color or another visual format carries a review decision, agree how that decision will appear in the exported data before handing it over.
Source: Microsoft's text and CSV saving guidance, checked November 17, 2026. The page lists Excel for Microsoft 365, Excel 2024 and Excel 2021; confirm the available format in the recipient's application.
Deliver the file with its limits
Return the agreed working file, a rule summary, a change record and unresolved questions. State whether duplicates were only flagged or actually removed with approval. Explain what checking occurred and what was outside scope.
Agree on revisions and retention before accepting the job. New records, a revised duplicate definition or changes to linked systems may be a separate task. Include communication, checking and handover in the proposed workload; this guide gives no market rate, demand estimate or income promise.
For broader work options, compare virtual assistant services and remote side hustles. A useful next step is one reversible practice file whose changes you can account for, rather than an offer to fix any spreadsheet regardless of its purpose or complexity.