Using Claude to Organize Messy Data: 2 Tips That Save Hours

Most “data problems” are not really analysis problems. They are formatting problems. Somebody exports a report, somebody else pastes it into an email, a third person types a few rows by hand, and now you have a file where the same customer is spelled four different ways and half the dates are text.

You can fix that by hand. It just costs you an afternoon every time.

Claude is unusually good at this particular chore, because cleaning and organizing data is mostly a language task: reading messy human input and deciding what it was supposed to mean. Here are two techniques I use constantly, and that you can put to work today.

Tip 1: Give Claude the target shape before you give it the mess

The most common mistake is pasting in a pile of data and saying “clean this up.” Claude will do something reasonable, but “reasonable” is not the same as “the exact layout my pivot table needs.”

Instead, describe the finished table first. Then hand over the mess.

I’m going to paste in a list of orders that came from three different systems. Convert it into a single table with exactly these columns, in this order:
CustomerName | OrderDate (YYYY-MM-DD) | ProductCode (uppercase) | Quantity (whole number) | Amount (number, no currency symbol) | Region

Rules:
– One row per order. No merged cells, no blank spacer rows, no totals row.
– If a date is ambiguous, use the US format and flag it.
– If a value is missing, put UNKNOWN rather than guessing.
– Return the result as CSV so I can paste it straight into Excel.

Here is the data:

Three things in that prompt do the heavy lifting. You named the columns, so Claude is not inventing a schema. You banned the things that break Excel later (totals rows, merged cells, blank rows). And you told it what to do when it is unsure, which is the difference between a clean file and a file with quiet, invented data in it.

That last rule matters more than people expect. Left to its own devices, any AI tool will try to be helpful and fill a gap. “Put UNKNOWN rather than guessing” turns a hidden risk into a visible one you can filter on.

Tip 2: Ask for the rules, not just the cleaned data

Here is the tip that actually saves the hours. After Claude hands you the clean table, ask this:

Before I use this, give me two things:
1. A change log — every transformation you applied, grouped by type, with one example of each.
2. An exceptions list — every row where you had to make a judgment call, with the original value, what you changed it to, and why.

Suddenly you are not trusting a black box. You are reviewing a short list of decisions. In my experience the exceptions list is usually 5 to 15 rows out of a thousand, and that is where every real error hides. Ten minutes of review on that list beats an hour of scrolling through the whole file.

There is a bonus. That change log is a specification. “Strip trailing spaces, uppercase product codes, map ACME/Acme Inc./ACME CORP to Acme Corporation, convert dates from DD/MM/YY” is now a documented, repeatable process. Next month, when the same ugly export lands in your inbox, you paste the change log back in and say “apply these same rules to this new file.” The second run takes two minutes instead of an afternoon.

Clean data once. Reuse the rules forever.

Want to see this done live?

I’m teaching a full session on exactly this workflow:

How to Use Claude for Analysis and Organizing Data
Thursday, September 10, 2026 — 11:00 AM Eastern
Live online, with time for your questions.

In that session I go well past these two tips. We work through real messy files, splitting and combining columns, reconciling two lists that should match but don’t, turning a wall of raw rows into a summary somebody will actually read, and getting Claude to explain what it found in the data instead of just reformatting it. You’ll leave with prompts you can use the same afternoon.

Full details and registration are here.

Plenty more on the calendar

Organizing data is one slice of what I teach. Over the next few months I’m also running live sessions on:

• Claude — using Claude in Excel, Claude for project management, Claude for HR professionals, Claude for small and mid-sized businesses, and Claude for non-technical professionals
• Microsoft Copilot — Copilot for Excel, Copilot across Microsoft 365 (Excel, Word, Outlook, Teams), Copilot for leaders, and Copilot for project management
• ChatGPT — ChatGPT for sales professionals, ChatGPT and Copilot together for daily office tasks, and ChatGPT for HR
• Head-to-head comparisons — Copilot vs. ChatGPT vs. Claude: which AI is best for business
• Excel and Power BI — pivot tables, Power BI for beginners and Excel users, and building reports people actually use

The full, current schedule lives on the front page of PCWebinars.com. New dates go up as they’re confirmed, so it’s worth a look even if nothing above is quite what you need.

About your trainer

I’m Tom Fragale. I’ve been a computer trainer and consultant for more than 30 years, and I’ve 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.

My whole approach comes down to one thing: you should be able to use what you learned before the end of the day. No theory for its own sake, no showing off features you’ll never touch. Just the handful of things that will save you real time, explained clearly, with your questions answered live.

Come join me on September 10 — or browse everything else at PCWebinars.com.