Claude in Excel: Two Ways to Fix a Messy Export - PC Webinars blog post

Claude in Excel: Two Ways to Fix a Messy Export

There is a spreadsheet on your machine right now that you rebuild by hand every month. I know, because I have watched a few thousand people do it in class. Somebody exports data out of the accounting system, it lands in Excel with the columns in the wrong order, the dates stored as text, and trailing spaces on every customer name. Then you spend ninety minutes cleaning it up before you can start the work you were actually asked to do. Learning to use Claude in Excel is the fastest way I have found to stop paying that ninety minutes over and over.

Claude does not replace knowing Excel. If you do not know what a lookup is, Claude will hand you one and you will not be able to tell whether it is right. What it does replace is the tedious middle part — the typing, the nested parentheses, the fourth trip to a forum thread from 2014 to find out why your formula returns #VALUE!. Here are two things I demonstrate in class that people use the same afternoon.

Claude in Excel: Describe the Cleanup Instead of Doing It

Anthropic now ships a Claude add-in that lives inside Excel itself. Go to the Home tab, click Add-ins, then Get Add-ins, and search for Claude. It opens as a task pane down the right-hand side of the window, reads the cells you point it at, and writes results back into the sheet.

The trick is to stop asking vague questions and start describing the finished state you want. Select your data range first, then type something like this:

This is a raw export. In the sheet named Raw, columns A through H, rows 2 to 4000: trim the extra spaces out of the customer names in column B, convert the text dates in column D into real Excel dates, split the full address in column F into separate city, state and ZIP columns, and put the cleaned version on a new tab called Clean. Leave the Raw tab untouched.

What comes back is not a paragraph of advice. It writes the new columns, applies the formulas, and tells you which cells it changed so you can spot-check them. This beats doing it by hand not just on speed — the same prompt works again on next month’s export, because you described the rule rather than clicking through this particular file.

One gotcha trips up nearly everyone: Claude only knows about the cells you actually give it. If you do not name the sheet and the range, it will guess, and it will guess wrong on a workbook with nine tabs. Always name the sheet, and always keep the raw export on a tab you tell it not to touch, so you have something to compare against when a number looks strange.

No add-in yet? Copy your header row plus about thirty representative rows, paste them into a chat at claude.ai, and ask for the formula. Thirty rows is enough for it to see the pattern without you pasting your whole customer list.

Ask for the Formula and the Reason Behind It

This is the habit that separates people who get real value out of AI from people who end up with a spreadsheet they are afraid to send to their boss. Never ask only for the formula. Ask for the formula, the explanation, and the failure mode, in one prompt:

Write an Excel formula that pulls the region from my Territory tab into column J of my Sales tab, matching on the rep's employee ID in column A. Use my actual column headers. Then explain in two sentences what each part of the formula does, and tell me the most likely reason it would return an error.

You get back a working formula, a plain-English breakdown, and a warning — usually something like “this will fail if the employee IDs are stored as text on one tab and as numbers on the other.” That last part is worth more than the formula. It is what a colleague would have told you if a colleague had been standing behind your chair.

The gotcha here is version drift. Claude will cheerfully hand you XLOOKUP or TEXTSPLIT, which are wonderful and which do not exist in Excel 2019. If you are not on Microsoft 365, say so: add “I am running Excel 2019, so use only functions available in that release.” It will fall back to INDEX and MATCH without complaint.

Plenty More on the Calendar

If you want to see all of this done live, on a real messy export rather than a tidy textbook example, I am running Using Claude in Excel on Wednesday, September 9. We cover connecting Claude to a workbook, the prompt wording that gets usable formulas on the first try, cleaning up an export end to end, and how to check the results before you trust them. You can ask questions as we go.

The full current schedule lives on the front page of PCWebinars.com, and new dates go up as they get confirmed. Register for this one, or browse the rest — there is usually something on Excel, Power BI, Access, Copilot or Claude within the next week or two.

About Your Trainer

I’m Tom Fragale. I have been a computer trainer and consultant for more than 30 years, and I have taught well over 30,000 people how to get more out of the software they already own — Excel, Access, Power BI, Word, Outlook, and now the AI tools that sit on top of all of it: Claude, ChatGPT, and Microsoft Copilot. My background is applied, not theoretical. Alongside teaching, I build real databases and reporting systems for clients in Excel, Access, and Power BI, so the examples I use in class are the problems I ran into last week, not textbook exercises. I also publish free tutorials on YouTube and longer courses online.

If any of this sounds like a problem you have been living with, come see the whole schedule and grab a seat at PCWebinars.com.

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *