{"course":{"slug":"ai-at-work","title":"AI at Work","url":"https://addaly.com/learn/ai-at-work","level":"beginner","track":"work"},"lesson":{"slug":"numbers-without-being-a-numbers-person","title":"Spreadsheets and data, without being a data person","url":"https://addaly.com/learn/ai-at-work/numbers-without-being-a-numbers-person","position":32,"of":73,"minutes":9,"answer":"Never ask AI for the answer to a calculation; ask for the formula, then test it on rows you can check by hand.","keyIdea":"Never ask AI for the answer to a calculation; ask for the formula, then test it on rows you can check by hand."},"body":{"format":"markdown","text":"There is a specific trap with AI and numbers, and once you see it you will never fall into it again.\n\n**Do not ask the model to do arithmetic. Ask it to write the thing that does the arithmetic.**\n\nPaste 300 rows of sales into a chat and ask for the total, and something is generating a plausible-looking number. Ask instead for the formula, run the formula yourself, and now a spreadsheet did the maths: a machine that has never once got a sum wrong. Same effort. Completely different reliability.\n\n## The formula you could not write\n\nThis is the highest-value use of AI for anyone who lives in spreadsheets but is not a spreadsheet expert. Describe the problem in your own words, including your actual column layout:\n\n> In Google Sheets. Column A is invoice date, column B is client name, column D is amount in rupees, column F says PAID or UNPAID. I want a formula in H2 that totals unpaid invoices for the client named in G2 that are more than 30 days old.\n\nYou get `=SUMIFS(D:D, B:B, G2, F:F, \"UNPAID\", A:A, \"<\"&TODAY()-30)` and an explanation of each part. You did not need to know SUMIFS existed. You do need to test it, which is the next section.\n\nThe same move works for pivot tables (\"how do I set this up to show monthly totals by branch\"), for cleaning (\"a formula to strip the leading apostrophe and trailing spaces from these phone numbers\"), and for the classic \"why does this return #VALUE!\": paste the formula and the error, and you will usually get the answer faster than searching.\n\n## Testing a formula in one minute\n\nNever trust a formula because it returned a number. Numbers always look right.\n\n1. **Run it on a handful of rows you can check by hand.** Ten rows, calculator, compare.\n2. **Try the awkward cases.** A blank cell. A zero. A date exactly 30 days old. A client name with a trailing space. These are where formulas break.\n3. **Sanity-check the scale.** If unpaid invoices come to ₹4,200 and you know it is roughly two lakh, stop.\n\nThat is a minute of work and it catches nearly everything.\n\n## Describing data, and the honest limit\n\nModern assistants can also take a spreadsheet file and analyse it: often by writing and running actual code behind the scenes, which is reliable arithmetic. Where they help:\n\n- **First look.** \"What is in this file? Any obviously broken columns, duplicates, impossible dates?\"\n- **Asking the question you cannot phrase.** \"Is there a pattern in which customers stop ordering?\" It will suggest angles, several of which will be worth checking.\n- **Charts.** \"Plot monthly revenue by region\" is far faster than clicking through chart menus.\n\nWhere it will mislead you: **it does not know your business.** It cannot know that April's spike was a one-off government order, that Branch 7 changed its recording method in June, or that \"customer\" means something different in two of your systems. It will find a pattern and explain it confidently. The explanation is a hypothesis dressed as a finding.\n\nAnd a hard line: it cannot tell you whether a difference is real or noise unless you ask properly, and \"sales are up 12% since the new packaging\" is not evidence that packaging caused it. If a decision worth real money rests on it, get someone who does statistics to look.\n\n## Where the accountability sits\n\nIf you paste a formula into the sheet that goes to the board, that formula is yours. \"The AI wrote it\" is not a sentence anyone will accept, and you would not accept it from a junior colleague either."},"check":{"question":"A shop owner pastes a year of transactions into a chatbot and asks for total revenue by month. She then pastes the same data in and asks for a formula, which she runs in her spreadsheet. The two sets of numbers differ slightly. Which is more likely correct, and why?","options":[{"text":"The spreadsheet, because the formula is executed arithmetic while the chat answer was generated text that resembles a calculation","correct":true},{"text":"They should be treated as equally likely, and she should reconcile them row by row to find the discrepancy","correct":false,"tempting":"Grants the generated numbers equal standing, which feels rigorous but wastes effort on output that was never a calculation."},{"text":"The chat answer, because the model reads every row directly while a formula may have the wrong range or miss rows","correct":false,"tempting":"Range errors are real, but they are cheap to test, whereas generated totals cannot be made reliable by any amount of checking."},{"text":"The spreadsheet, because chatbots cannot process a year of data and will have silently truncated the paste","correct":false,"tempting":"Truncation may well happen, but attributing the difference to volume misses that even ten rows of chat arithmetic is unreliable in kind."}]},"prev":{"slug":"what-the-columns-actually-mean","title":"Before you analyse it, find out what you have","url":"https://addaly.com/learn/ai-at-work/what-the-columns-actually-mean"},"next":{"slug":"cleaning-a-messy-spreadsheet","title":"Cleaning a spreadsheet somebody else built","url":"https://addaly.com/learn/ai-at-work/cleaning-a-messy-spreadsheet"},"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'Spreadsheets and data, without being a data person', in AI at Work. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/ai-at-work/numbers-without-being-a-numbers-person (Accessed: 31 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"Spreadsheets and data, without being a data person","courseTitle":"AI at Work","module":{"index":4,"slug":"working-from-your-own-material","title":"Working from your own material"},"answer":null,"keyIdea":"Never ask AI for the answer to a calculation; ask for the formula, then test it on rows you can check by hand.","hero":{"kind":"plate","image":null,"alt":null,"caption":null},"blocks":[{"t":"p","runs":[{"text":"There is a specific trap with AI and numbers, and once you see it you will never fall into it again."}],"lede":true},{"t":"p","runs":[{"text":"Do not ask the model to do arithmetic. Ask it to write the thing that does the arithmetic.","b":true}]},{"t":"p","runs":[{"text":"Paste 300 rows of sales into a chat and ask for the total, and something is generating a plausible-looking number. Ask instead for the formula, run the formula yourself, and now a spreadsheet did the maths: a machine that has never once got a sum wrong. Same effort. Completely different reliability."}]},{"t":"h2","id":"the-formula-you-could-not-write","runs":[{"text":"The formula you could not write"}]},{"t":"p","runs":[{"text":"This is the highest-value use of AI for anyone who lives in spreadsheets but is not a spreadsheet expert. Describe the problem in your own words, including your actual column layout:"}]},{"t":"quote","runs":[{"text":"In Google Sheets. Column A is invoice date, column B is client name, column D is amount in rupees, column F says PAID or UNPAID. I want a formula in H2 that totals unpaid invoices for the client named in G2 that are more than 30 days old."}]},{"t":"p","runs":[{"text":"You get "},{"text":"=SUMIFS(D:D, B:B, G2, F:F, \"UNPAID\", A:A, \"<\"&TODAY()-30)","code":true},{"text":" and an explanation of each part. You did not need to know SUMIFS existed. You do need to test it, which is the next section."}]},{"t":"p","runs":[{"text":"The same move works for pivot tables (“how do I set this up to show monthly totals by branch”), for cleaning (“a formula to strip the leading apostrophe and trailing spaces from these phone numbers”), and for the classic “why does this return #VALUE!”: paste the formula and the error, and you will usually get the answer faster than searching."}]},{"t":"h2","id":"testing-a-formula-in-one-minute","runs":[{"text":"Testing a formula in one minute"}]},{"t":"p","runs":[{"text":"Never trust a formula because it returned a number. Numbers always look right."}]},{"t":"ol","start":1,"items":[[{"text":"Run it on a handful of rows you can check by hand.","b":true},{"text":" Ten rows, calculator, compare."}],[{"text":"Try the awkward cases.","b":true},{"text":" A blank cell. A zero. A date exactly 30 days old. A client name with a trailing space. These are where formulas break."}],[{"text":"Sanity-check the scale.","b":true},{"text":" If unpaid invoices come to ₹4,200 and you know it is roughly two lakh, stop."}]]},{"t":"p","runs":[{"text":"That is a minute of work and it catches nearly everything."}]},{"t":"h2","id":"describing-data-and-the-honest-limit","runs":[{"text":"Describing data, and the honest limit"}]},{"t":"p","runs":[{"text":"Modern assistants can also take a spreadsheet file and analyse it: often by writing and running actual code behind the scenes, which is reliable arithmetic. Where they help:"}]},{"t":"ul","items":[[{"text":"First look.","b":true},{"text":" “What is in this file? Any obviously broken columns, duplicates, impossible dates?”"}],[{"text":"Asking the question you cannot phrase.","b":true},{"text":" “Is there a pattern in which customers stop ordering?” It will suggest angles, several of which will be worth checking."}],[{"text":"Charts.","b":true},{"text":" “Plot monthly revenue by region” is far faster than clicking through chart menus."}]]},{"t":"p","runs":[{"text":"Where it will mislead you: "},{"text":"it does not know your business.","b":true},{"text":" It cannot know that April’s spike was a one-off government order, that Branch 7 changed its recording method in June, or that “customer” means something different in two of your systems. It will find a pattern and explain it confidently. The explanation is a hypothesis dressed as a finding."}]},{"t":"p","runs":[{"text":"And a hard line: it cannot tell you whether a difference is real or noise unless you ask properly, and “sales are up 12% since the new packaging” is not evidence that packaging caused it. If a decision worth real money rests on it, get someone who does statistics to look."}]},{"t":"h2","id":"where-the-accountability-sits","runs":[{"text":"Where the accountability sits"}]},{"t":"p","runs":[{"text":"If you paste a formula into the sheet that goes to the board, that formula is yours. “The AI wrote it” is not a sentence anyone will accept, and you would not accept it from a junior colleague either."}]}],"check":{"question":[{"text":"A shop owner pastes a year of transactions into a chatbot and asks for total revenue by month. She then pastes the same data in and asks for a formula, which she runs in her spreadsheet. The two sets of numbers differ slightly. Which is more likely correct, and why?"}],"options":[{"text":[{"text":"The spreadsheet, because the formula is executed arithmetic while the chat answer was generated text that resembles a calculation"}],"correct":true,"tempting":null},{"text":[{"text":"They should be treated as equally likely, and she should reconcile them row by row to find the discrepancy"}],"correct":false,"tempting":[{"text":"Grants the generated numbers equal standing, which feels rigorous but wastes effort on output that was never a calculation."}]},{"text":[{"text":"The chat answer, because the model reads every row directly while a formula may have the wrong range or miss rows"}],"correct":false,"tempting":[{"text":"Range errors are real, but they are cheap to test, whereas generated totals cannot be made reliable by any amount of checking."}]},{"text":[{"text":"The spreadsheet, because chatbots cannot process a year of data and will have silently truncated the paste"}],"correct":false,"tempting":[{"text":"Truncation may well happen, but attributing the difference to volume misses that even ten rows of chat arithmetic is unreliable in kind."}]}],"order":[1,2,3,0],"why":null},"prev":{"slug":"what-the-columns-actually-mean","title":"Before you analyse it, find out what you have","minutes":9},"next":{"slug":"cleaning-a-messy-spreadsheet","title":"Cleaning a spreadsheet somebody else built","minutes":10}}}