For Environmental Scientist / Sustainability Analysts ·
What you'll accomplish
A comparison column that is wrong in one cell, or an average that quietly includes a reporting limit, is the kind of slip that survives three reviews because the sheet looks tidy. In this guide you upload a small practice workbook with two planted errors, ask ChatGPT to recompute the columns from the raw data, and compare its answer with your sheet cell by cell. The sheet stays the record. ChatGPT only points at cells you then check yourself.
What you'll need
Data caution. Upload only invented or public files here. Real monitoring workbooks, lab reports, site codes tied to a client, addresses and photos stay in the firm's approved systems unless the firm and the client approve ChatGPT for that data. Before you upload anything non-public, open ChatGPT's data controls and check whether chats and uploaded files are used to train models and how long they are kept. When you share a sheet, use column names and codes such as "Site S-01" and "MW-3" and strip anything that identifies a client.
Name the file s01-practice.xlsx. All values are invented. The screening value 50 is an invented screening value, not a real standard.
Tab "Results". Row 1 holds headers: A Location, B Sample date, C Result (ug/L), D Qualifier, E Screening value, F Comparison.
| Row | A | B | C | D | E |
|---|---|---|---|---|---|
| 2 | MW-1 | 2026-09-14 | 12 | 50 | |
| 3 | MW-2 | 2026-09-14 | 68 | 50 | |
| 4 | MW-3 | 2026-09-15 | 5 | ND | 50 |
| 5 | MW-4 | 2026-09-15 | 50 | 50 | |
| 6 | MW-5 | 2026-09-15 | 140 | 50 |
ND means not detected, and the result cell holds the reporting limit.
=IF(OR(A2="",C2="",E2=""),"",IF(D2="ND","Not detected",IF(C2>E2,"Above SV","At or below SV")))
s01-practice.xlsx.What you should see: The file name shown as an attachment on your message. Troubleshooting: If no upload control appears, your plan or account type may not include file uploads. Check OpenAI's help pages on data analysis and file uploads.
Paste this prompt with the file attached.
This workbook is invented practice data for Site S-01. Do not treat any value as real.
On the Results tab, ignore columns F and H to I. Using only columns A to E, recompute for each row the comparison text: "Not detected" if D is ND, "Above SV" if C is greater than E, otherwise "At or below SV". Show a table with the row, my sheet's value in F, your recomputed value, and a column that says SAME or DIFFERENT.
Then list the cells that differ and say in one line what formula logic could explain each difference. Do not edit the file.
Next, recompute the three counts and the average of detected results (exclude rows where D is ND). Show each calculation step. Compare with cells I2 to I5 and list any that differ.
Finally, make a clustered bar chart of the Analyte A result by location for June and September, using the June tab and the Results tab. Mark MW-3 as ND in the chart title or a note, since its value is a reporting limit.
Label everything as a draft for my review.
What you should see: A comparison table with one DIFFERENT row (MW-4, row 5), a list of differing summary cells, and a chart. If ChatGPT says it cannot read the file, re-save it as .xlsx or .csv. OpenAI's data analysis page lists spreadsheet formats it works with, and descriptive headers with one record per row help.
Do this before you accept anything. You know the answers, so use them as the test of the tool.
| Item | Your sheet showed | Correct value |
|---|---|---|
| F2 (MW-1) | At or below SV | At or below SV |
| F3 (MW-2) | Above SV | Above SV |
| F4 (MW-3) | Not detected | Not detected |
| F5 (MW-4) | Above SV | At or below SV (50 is not above 50) |
| F6 (MW-5) | Above SV | Above SV |
| I2 count Above SV | 3 | 2 |
| I3 count At or below SV | 1 | 2 |
| I4 count Not detected | 1 | 1 |
| I5 average | 55 | 67.5 |
The 55 comes from 275 divided by 5, which includes the reporting limit 5 for MW-3. The correct 67.5 comes from 12 plus 68 plus 50 plus 140, which is 270, divided by 4 detected results.
If ChatGPT missed either planted error, or invented a third one, note that. The tool's answer is a lead, never a result.
>= to >, so it matches F2, F3, F4 and F6.=AVERAGEIF(D2:D6,"<>ND",C2:C6)
What you should see: All five comparison cells and all four summary cells matching the correct values above.
ChatGPT's chart is a draft picture. Confirm each bar against the sheet. June values are 15, 55, 5 (ND), 47 and 128. September values are 12, 68, 5 (ND), 50 and 140. If a bar differs, the chart is wrong, not the sheet.
Change the row numbers to match each row.
2. In F5 paste the version with the planted error (a), which uses >= where it should use >:
=IF(OR(A5="",C5="",E5=""),"",IF(D5="ND","Not detected",IF(C5>=E5,"Above SV","At or below SV")))
=COUNTIF(F2:F6,"Above SV")
=COUNTIF(F2:F6,"At or below SV")
=COUNTIF(F2:F6,"Not detected")
=AVERAGE(C2:C6)
What you should see: F5 shows "Above SV" for the value 50. I2 shows 3, I3 shows 1, I4 shows 1, and I5 shows 55.
Tab "June". Same columns A to D, with invented values and an invented sample date of 2026-06-16 in column B: MW-1 15, MW-2 55, MW-3 5 with qualifier ND, MW-4 47, MW-5 128.
Troubleshooting: If F5 does not show "Above SV", recheck that you typed >= and not >.