How to import a CSV into Excel and verify the result
Start in a blank workbook, open Excel's text or CSV import flow, and preview the file before loading it. Choose the field delimiter and encoding, preserve identifier columns as text, and convert amounts and dates according to the source's conventions. Keep the original CSV. Verify the resulting worksheet with a small set of known values and a control total before using it in a report or another import.
In this guide
Establish what a correct import must contain
Imagine a studio manager receiving a monthly expense export. It has reference, posting date, supplier, amount and category columns. Some supplier names contain accents. References begin with zeros. Returns appear as negative amounts. A spreadsheet that opens without an error can still alter all three features. Write down your expectations from the source file or accompanying statement so you can test more than whether there is a visible table.
Pick control records deliberately
Choose an ordinary expense, a negative refund, a reference starting with zero, a supplier name containing punctuation, and one record from the final page or final day. Pick at least one date whose day exceeds twelve if you need to establish a date convention. If a file contains no such date, use the issuer's specification. A random group of five morning café purchases may look reassuring while failing to cover the actual risks.
Preserve an independent reference
Make a separate working copy of the CSV and use a new workbook for the import. Keep any original PDF that explains totals or document scope. When a discrepancy appears you will need to determine whether it came from the source, the earlier extraction, or Excel. Opening a CSV, making changes, and saving it over the original removes one of the most valuable comparison points. Choose filenames that distinguish received data from checked output.
Inspect the preview in the right order
Find the data import command for a text or CSV file in your Excel version. Where a transformation editor is available, enter it before loading. Microsoft describes its data import approach in its guide to preserving numerical text during import. For environments with the legacy import interface, the Text Import Wizard documentation explains delimiter, qualifier, and column formatting choices.
First make the columns appear
Check the header and confirm the expected number of fields. A wide cell containing a complete record suggests a delimiter mismatch. A surname or product name split at a comma suggests an issue with field quoting or the export itself. If only particular rows have too many fields, compare them in a text editor with one successful row. Change the delimiter only when it matches the evidence in the file.
Then make the characters readable
Inspect an accented supplier name and a currency symbol if one appears in the source. Garbled characters can point to a character encoding mismatch. Select the encoding documented for the file and check the preview again. Replacing garbled names with similar-looking spellings can create duplicate vendors later. If the text already looks damaged in the original source, changing the spreadsheet's import settings may not recover it. Ask for a fresh export when necessary.
Classify fields before converting them
Identifiers, names and category labels generally belong in text columns. Amounts used for arithmetic need a verified numeric interpretation. Dates need the source day and month order. In an editor that automatically detects types, examine the applied steps. A type conversion may have already removed a leading zero before you see the editing grid. Remove or correct that conversion and set the correct types while the original text is still available.
Why row counts alone mislead
A file can contain the expected 246 expense rows and still be wrong because every amount is stored as text or an account code has been shortened. Examine the fields you intend to calculate with and the fields you intend to match on separately. Sum a verified amount column and perform a lookup on a known identifier. These tests exercise different properties of the import, and passing one gives little assurance about the other.
Inspect text that resembles a formula
A description may start with punctuation used in spreadsheet expressions. For text columns from another person or system, choose a controlled import and inspect how the destination displays those values. Treat unexpected calculations in a description or identifier column as a reason to stop and check the workflow. Preserve the original vendor wording as text. If the recipient uses another application, test its own behavior with harmless fictional descriptions rather than assuming that Excel's import choices apply everywhere.
Walk through a small expense example
Suppose three expense records have the original amounts 24.50, 18.00 and −6.50. If this is a simple signed expense list, the expected net expense is 36.00. The negative item is a refund. Check both the net and the individual refund. If the workbook total is 42.50, a missing refund or a refund stored as ignored text may explain it. If the total is 49.00, the refund might have been converted to a positive expense.
Follow a mismatch to a specific stage
Find the refund by its unique source reference. Compare its visible amount in the PDF, its literal field in the CSV, and its underlying value in the workbook. A Unicode minus from a PDF may behave differently from an ordinary hyphen in numeric parsing. Accounting parentheses can express a negative amount under an agreed source convention. Convert only the affected amount field after confirming the convention, then repeat the three-record calculation.
A tax-inclusive invoice total or month-end summary line can double the sum if it becomes an ordinary expense row. Identify what a row represents in this file. If it contains transactions plus a summary, keep the summary outside the amount column used for arithmetic. If the total still disagrees, test separate categories or statement dates to narrow the search. A source period ending at midnight may differ from a report grouped by posting date.
Load a worksheet you can work with
Name the worksheet clearly and save the working result as a workbook if it will contain formulas, notes or several tabs. A CSV represents a text table. Keeping the analysis in a workbook avoids using a text export for features it cannot carry. Keep any untouched source table separate from calculated columns so a colleague can distinguish original values from classifications or adjustments added during cleanup.
Add a tiny verification sheet
Record the source filename, expected data row count, expected total, checked references and any exclusions such as a printed subtotal. Say how you determined the expected total. Copying it from the very worksheet you are testing gives no independent evidence. A PDF footer, a supplier control total, or arithmetic on an independently selected sample provides a stronger basis. Keep the verification sheet practical enough that another colleague can repeat the checks.
Decide what should happen on refresh
A reusable import can be helpful when the same supplier delivers a stable monthly layout. Document where the input file lives and which worksheet shows the imported results. Keep manual annotations in a separate keyed table when refreshing might replace imported rows. If a supplier renames columns or changes the date format, check the preview and control values again. An unchanged filename is no guarantee that the underlying layout stayed the same.
Send the right output to the next person
Ask whether the recipient wants a workbook they can read or a CSV that another program will ingest. For a workbook, reopen the saved file and confirm the intended tables and checks are accessible. For CSV delivery, export the intended sheet only and inspect the final text file. A recipient may need the agreed delimiter, named headers, and specific date representation. Recheck one code with leading zeros and one negative amount after export.
When the source began as a PDF
If the import preview reveals duplicate page headings, wrapped descriptions or a line from a printed footer, return to the PDF to CSV extraction workspace and compare the offending rows with the PDF. Fixing an extraction issue before it becomes a repeatable import rule can prevent unnecessary per-month cleanup. For a digital bank statement intended specifically for Excel, the statement workbook converter offers an alternative output format.
A final handover question
Could the next person explain why the amount total is correct and how to recover the unmodified input? If the answer requires your memory, add a note containing the file period, included rows and the checks you performed. If totals agree but a high-value transaction lacks its original identifier, flag that exception separately. Accurate arithmetic and reliable record identity are both necessary for a workbook that someone else can use with confidence.
Investigate a lookup that still fails
Suppose the expense total agrees but a supplier code never matches the master list. Compare the text from the imported cell with the exact master value. Look for an unseen trailing space, a different dash character or a prefix removed by an earlier system. Confirm which characters are significant before trimming or replacing them across an entire column. A visible gap between two groups of characters may be part of a standardized code, while a trailing space copied from a fixed-width layout may be harmless padding.
Create a separate comparison key if normalization is necessary. Leave the source value beside it and document the normalization rule. Test a pair of distinct supplier codes that look similar before approving the rule. If both collapse into one key, the transformation is too aggressive for that dataset. This kind of check belongs alongside row counts and amount totals, especially when the workbook will feed a supplier-payment review, reconciliation or monthly purchasing analysis.