A clean-looking column can still contain the wrong information. A few deliberate checks protect the meaning behind the cells.
By Arbab Naseebullah Kasi, Chief Executive Officer - CEO, Feel Worldwide Foundation Inc.
The file opens. The columns line up. Nothing flashes red. For someone trying to finish a busy afternoon's work, that can feel like confirmation that the information has arrived safely.
Sometimes it has not.
A spreadsheet can make an interpretation before its user has made a decision. Something that looks like a number may become a number, even when it was meant to remain an identifier. A string that resembles a date may acquire a date's meaning. The result can look tidier than the original while becoming less faithful to it.
This is not a reason to abandon a useful tool. It is a reason to treat importing information as a small act of translation. The question is not only whether the file opened. It is whether the values still mean what their source intended.
A label is not always a quantity
Consider a fictional equipment register containing the asset code 00427. The code identifies one item. It is not a quantity to add to another item, and its opening zeros may be part of the agreed naming system.
If a spreadsheet turns it into 427, the cell may still look perfectly reasonable. Yet it no longer matches the original identifier character for character. A later search or comparison could fail because two systems are no longer using the same label.
Microsoft documents that Excel can remove leading zeros and has a numeric precision limit of 15 significant digits. Long identifiers therefore need text handling. Behaviour depends on the software and import settings. Microsoft's guidance on leading zeros and large numbers.
A useful first distinction is therefore conceptual: what is this column for? An identifier names something. A quantity measures or counts something. Both may contain digits, but they need not receive the same treatment.
In the fictional register, the organization should preserve the asset code according to its source rules, while storing the number of replacement parts as a quantity suitable for arithmetic. Making every column text is not the goal. Choosing deliberately is.
The problem has reached published research
This is more than a hypothetical office inconvenience. In a 2016 study, Mark Ziemann, Yotam Eren and Assam El-Osta examined supplementary spreadsheet material from 18 scientific journals covering 2005 through 2015. They confirmed gene-name errors in 704 articles, representing 19.6 percent of the selected papers with Excel gene lists. Some gene symbols had been interpreted as dates or other numeric forms. The original Genome Biology paper.
Those figures describe that study's selected scientific literature and historical period. They are not an error rate for all spreadsheets, all researchers or today's software.
The lesson is still useful outside genomics. A value can change without becoming visibly absurd. A date looks like a legitimate thing to find in a cell. Detecting the mistake requires knowing that this particular column was supposed to contain something else.
That makes familiarity with the work as important as familiarity with the software. The person who understands what a code represents may notice a problem that a technically confident colleague overlooks.
Agree on meaning before choosing a format
Before importing a recurring file, write a short explanation of its important fields. It need not become a lengthy manual. A handful of clear definitions can prevent a great deal of guesswork.
For a fictional community equipment programme, the explanation might say: asset codes retain all characters; checkout dates describe when an item left the store; return dates stay blank until a return is recorded; item counts are whole-number quantities. It should also say whether each row represents an item, a checkout or a person.
That last distinction matters. Ten rows do not necessarily mean ten people. One person might appear several times because they borrowed several items. A correctly imported file can still produce a misleading conclusion if the reader misunderstands what a row represents.
Clarify missing information too. A blank return date may mean the item has not returned, the date is unknown or the record is incomplete. Those are different states. The team should decide how to represent them rather than ask a later reader to infer the answer from an empty cell.
These definitions belong with the workflow, where a substitute colleague can find them. Knowledge that exists only in the original author's memory is easily lost when a routine task changes hands.
Keep the source and make the import deliberate
Retain an unchanged source copy in an appropriate, access-controlled location before transforming a working copy. This creates a point of comparison. It does not mean keeping personal information forever or creating uncontrolled duplicates; use the organization's retention and access rules.
Where the software offers an import preview, use it to inspect how fields will be interpreted. Check the actual tool and version rather than relying on a remembered sequence of clicks from another computer.
Microsoft describes using Excel's Get & Transform tools to assign text types to relevant imported columns. It also identifies Automatic Data Conversions controls in Microsoft 365 and Excel 2024 editions for behaviours such as removing opening zeros or interpreting certain strings as dates. Availability depends on the version. Microsoft's import guidance.
Changing how a damaged value looks afterward is not a substitute for recovering the original. If information has already been lost, return to the source and import again with the appropriate interpretation. Do not guess missing characters because a pattern appears obvious.
The aim is a repeatable route from source to working data. A colleague should be able to explain what changed intentionally and what was preserved.
Test the awkward cases, not only the first row
A quick glance at ordinary entries can miss the values most likely to expose a problem. Build a small, invented test file containing the kinds of entries your real workflow must preserve.
For example, a test might include an asset code beginning with zeros, a longer identifier, an explicitly missing value and an ordinary quantity. Use fictional material so the exercise does not expose anyone's personal records.
Write down the expected result before opening the file. If the identifier is supposed to remain 00427, the check is an exact match with 00427, not a judgment that 427 looks close enough.
After importing real data, compare more than appearance. Check the number of records, important field names and a purposeful sample of values against the source. Review unusually long identifiers, empty fields and categories that would be easy to confuse. Where a total is meaningful, compare it with an independently established source total.
No single check proves that everything is correct. A matching row count does not show that every identifier survived, and a matching total does not establish that the values were attached to the right records. Choose several checks that address the ways this particular file could go wrong.
For consequential decisions, arrange an appropriate independent review. The more someone could lose from a mistaken record, the less reasonable it is to rely on a quick visual scan.
Make corrections explainable
If a check finds a discrepancy, pause before distributing the working file. Identify whether the issue came from the source, the import or a later edit. Keep those possibilities separate.
Suppose the fictional equipment register contains two codes that differ only by an opening zero. A colleague should not automatically merge them as duplicates. The right next step is to ask the record owner whether the codes identify different items or reflect an inconsistency in the source.
Record the decision once it is resolved. Note what was corrected, why and which source supported it. Avoid silently replacing an uncertain value with the answer that makes the sheet look most complete.
If a flawed version has already been used, tell the relevant colleagues which version is affected and provide a clear replacement. The useful message is specific: which field changed, what decisions may need rechecking and where the corrected file is available. Blame contributes less than a dependable recovery process.
Help the next person understand the file
The UK government's spreadsheet guidance, updated in October 2023, recommends a cover page with information such as purpose, publication date and version. It also recommends consistent column naming and warns that software limits can cause information to be lost or omitted. These are practical design principles, not a guarantee that a documented workbook is error-free. Government guidance on creating and sharing spreadsheets.
For a small team, a brief opening note can explain the source, the date imported, the intended use and the checks completed. State important limitations plainly. “Attendance records received through Friday” is more informative than “complete report” when later submissions are still possible.
Name the responsible person or role so questions have somewhere to go. If the file is only an intermediate working copy, say so. Presentation should help readers understand the status of the information, not lend unfinished work the appearance of final authority.
The next person may be a colleague tomorrow, a new volunteer next month or you after enough time has passed to forget the details. Clear explanations are a form of kindness to all three.
A small pause can protect a larger decision
The most useful improvement may take place in the minute between opening a file and trusting it. Ask what the important columns mean, preserve the source and check a few values that would reveal a mistaken interpretation.
This is not glamorous work. It rarely appears in a launch announcement. Yet dependable records support the ordinary promises organizations make: to find the right item, recognize the right person and report what actually happened.
Feel Worldwide Foundation Inc. welcomes practical partnerships that help communities and organizations strengthen their capabilities. To explore collaboration or support the Foundation through a donation, visit the Foundation's official website.
Which column in your most-used spreadsheet deserves a closer look before the next decision depends on it?

