Teach CSV Import Checks with a Tiny Fictional Transaction Table

A spreadsheet can make an imported table look convincing before anyone has checked what the cells mean. A date looks like a date, a total has two decimal places, and a short reference number seems harmless. Yet automatic type detection may have changed all three. This classroom exercise teaches learners to inspect those changes using a tiny fictional transaction table. It does not require a real account, a genuine bank statement, a paid service, or financial advice. The learning goal is reproducible data handling, not a conclusion about anybody's finances.

Disclosure and boundaries

I am associated with Bank Statement Converter, the optional resource described below. This original lesson was drafted with AI assistance and reviewed for its example arithmetic and product statements; it is not a report of a measured conversion benchmark. Every transaction reference and amount in the exercise is invented. Do not turn the worksheet into an imitation official statement or add a real bank's logo, signature, or customer details. A clearly labelled teaching table is sufficient for investigating spreadsheet behavior without exposing someone else's records.

Define the lesson before choosing software

Ask students to explain four things at the end: which fields should remain text, how an ambiguous date was handled, why a sum does not prove row accuracy, and how another reader can repeat the import. None of those objectives depends on a particular spreadsheet brand. Prepare the same source table for all groups and ask them to record their own import settings. Differences between groups are useful observations, not a competition to claim one application is universally better. A correct explanation matters more than a pretty screenshot.

Build a deliberately small source table

Use these fictional fields: source row, date as written, reference, description, money out, and money in. Give the first row the teaching reference 000042 and the amount 999.90 in money out. Give the next row reference 000043 and 0.10 in money out. Leave money in blank for both. Label the table SYNTHETIC CLASSROOM DATA in an accompanying note. These are generic teaching references, not account or routing numbers. A small table makes every transformation inspectable and keeps discussion focused on evidence instead of data volume.

Record the source separately from the working copy

Keep an unchanged copy of the fictional source and create a working import for each group. Give each working file a descriptive version name and record what was changed. A source row identifier should survive sorting and filtering, because row position in the current spreadsheet is not a durable identifier. If a student corrects an interpretation, preserve the original value in one column and the reviewed interpretation in another. The class should be able to discuss a decision without having to reconstruct a silently overwritten source.

Show why identifiers are not quantities

Reference 000042 is text even though it contains only digits. A spreadsheet might display it as 42 after automatic conversion. Those two strings can mean different things to a downstream matching process. Ask learners to import the reference column explicitly as text, then close and reopen a disposable export to check whether the zeros survived. Formatting a converted number to display extra zeros can mask the original loss rather than recover it. The important question is what data the file actually retains, not just what the current cell displays.

Leave the ambiguous date unresolved until evidence exists

Add the fictional date string 03/04/2026 to a test row. It can be interpreted differently under different regional settings. Keep the original string intact and add an interpretation status instead of guessing a month and day. If the exercise provides an explicit source convention, students can document that convention and create a normalized date in a separate column. If no convention is supplied, unresolved is the correct result. A lesson that rewards guessing teaches the wrong habit, even when the spreadsheet happens to show the instructor's preferred answer.

Separate numerical meaning from appearance

The two outgoing amounts total 1,000.00 in the fictional example. As integer cents, the same check is 99,990 plus 10 equals 100,000. Students can use the integer version to discuss units and avoid confusing a displayed decimal with a string containing punctuation. Different applications and formulas handle decimal arithmetic differently, so do not claim that one example proves all calculations exact. State the rounding convention used by the exercise and keep it separate from any real institution's reporting or accounting rules.

Understand blanks before filling them

A blank money-in cell might mean no incoming movement in this teaching table, but a blank imported field can also mean unavailable or not yet reviewed. Define the exercise's convention in a short data dictionary. Do not automatically fill every blank with zero merely because it makes a formula easier. A downstream reader should be able to distinguish a known zero from missing information. Have students add a review-status column rather than editing the raw table whenever they encounter a value whose meaning has not been established.

Make descriptions test the file format

Add a harmless description such as Sample materials, classroom only. The comma makes it useful for demonstrating how CSV quoting works. Then try an explicitly invented description containing quotation marks or a line break in a disposable copy. Reopen the file and check whether the description remains in one logical record. Counting visible spreadsheet lines is not always the same as counting records in a correctly quoted file. The class should inspect the import result, not conclude that every comma or new line is automatically a formatting error.

Keep imported text from becoming an instruction

Unexpected source descriptions should be treated as data. Configure the import so that a text value resembling a spreadsheet formula is not executed as a formula. This lesson does not need a working attack string or a demonstration involving external connections. The safe observation is simply that a value intended as text must remain text. Ask learners to document the relevant import setting or sanitization policy and retain the unchanged source separately. A setting that changes visible text should itself be recorded as a transformation, not hidden from the reviewer.

Use a converter only for the preparation step

When an authorized source document is a PDF or image, Bank Statement Converter is one optional way to prepare a structured table. Its official site, https://bank-statement-converter.online/, describes Excel, CSV, and JSON outputs, individual and batch conversion, and separate, ZIP, or merged downloads. Current public pricing lists paid subscriptions and permanent page-credit packs. This classroom lesson does not require buying credits or uploading financial documents. Conversion can reduce retyping, but it does not establish a document's authenticity, completeness, or suitability for an accounting decision.

Official product illustration, not a classroom result, statement example, or claim about conversion accuracy.

Check rows even when the total matches

A matching total can conceal offsetting mistakes. In a separate test copy, imagine changing one value upward and another downward by the same amount. The sum can remain unchanged while both rows are wrong. Students should compare row count, source references, dates, descriptions, and amounts as well as the total. An expected total is one check, not a replacement for inspection. Keep deliberate test changes clearly labelled and separate from the reviewed table so that a teaching counterexample cannot accidentally become the final version.

Distinguish duplicates from similar wording

Copy one synthetic row into a disposable exercise file and ask another group to identify it. Identical descriptions alone are insufficient evidence of duplication; legitimate records can share a description or amount. Compare the source row identifier and the source table before deciding what happened. Do not delete a row simply because the resulting total looks more convenient. The useful output is a short explanation of the evidence for the duplicate and the location of the unchanged source, rather than a worksheet with one unexplained missing line.

Perform a round trip instead of trusting the first view

Save a disposable CSV, close it, and reopen it using documented settings. Compare the leading zeros, date strings, number of records, quoted descriptions, blanks, and amounts. Keep a separate workbook if the lesson needs formulas, comments, review colors, or multiple tabs, because a CSV does not retain all workbook features. A successful first import is not proof that a later export will preserve the same choices. The round trip reveals which parts of the workflow depend on application defaults that another reader may not share.

Check character encoding explicitly

Include an invented description with an accented character and confirm that it survives the round trip. Record the selected character encoding rather than accepting a garbled display as a harmless cosmetic difference. A changed character can interfere with later matching or make two descriptions appear unrelated. Keep this test independent of the amount check: a correct sum says nothing about the preservation of text. If two applications disagree, compare the unchanged file and their import settings before editing the description to make it look familiar.

Evaluate the explanation, not the confidence

A brief handoff should name the source version, import settings, unresolved fields, corrections, and checks performed. Ask students to include one limitation they could not resolve from the supplied evidence. They should not claim to have validated a real customer balance, detected fraud, or produced a professional reconciliation. The exercise demonstrates data literacy on invented rows. If a real task involves financial reporting, tax, legal interpretation, or a disputed payment, appropriate qualified review is a different requirement and cannot be replaced by this classroom checklist.

End with a repeatable result

The lesson is complete when another group can reopen the same fictional source, reproduce the documented interpretation, and identify which questions remain unanswered. Keep the original table, data dictionary, working file, and short review note together. If the groups disagree, locate the exact setting or source ambiguity behind the disagreement before changing the answer. A visible, well-explained uncertainty is a better learning result than a silently corrected spreadsheet whose neat appearance disguises the decisions that made it look consistent.