Claude for Variance Analysis: Two Ways to Explain Budget Gaps

Month-end hits and the same hour disappears. You pull the budget-versus-actual export, scan down the column looking for the numbers that moved, then sit there trying to write two sentences explaining why travel is over by eleven thousand dollars. The math takes five minutes. The explaining takes the rest of the morning. That second part is where Claude for variance analysis earns its keep.

I have watched people do this in class for years, and the frustrating part is that the analysis is already sitting in the spreadsheet — nobody has time to dig it out line by line. Claude does not invent the numbers. It reads the table you already have and tells you which handful of rows actually explain the total.

Two things I show people. Both take about ten minutes, and neither requires changing how your workbook is built.

Using Claude for Variance Analysis on a Budget-vs-Actual Table

Start with the simplest move: select the range in Excel, press Ctrl+C, and paste it straight into the Claude chat box. Excel puts tab-separated text on the clipboard, so the columns survive the paste — no uploading, no converting to CSV. For a big sheet, use the paperclip icon and attach the .xlsx instead.

Then ask for the analysis in the shape you need it. This is the wording I use:

Here is a budget vs actual table by GL account. Calculate the dollar variance and percent variance for each row. Then tell me the five rows that account for the largest share of the total unfavorable variance, with the cumulative percent of the total they represent. For each one, write a one-sentence plain-English commentary a non-finance manager would understand. Do not comment on rows you cannot explain from the data.

What comes back is the part you were going to do by hand: a ranked list, the cumulative percentage so you know where to stop reading, and draft commentary you can edit rather than write. The ranking matters more than it sounds — a twenty-row report usually has three rows that explain eighty percent of the gap, and the manual approach reads all twenty anyway.

One gotcha: delete any title rows above your real header row before you copy. A merged “FY26 Operating Budget” banner sitting over the columns throws the mapping off, and you get a confident answer built on the wrong headers. And that last line in the prompt is doing real work — without it you get plausible-sounding reasons for variances the spreadsheet contains no evidence for. With it, you get an honest “cannot determine from the data provided,” which is what you want on a row you need to go ask somebody about.

Turn the Answer Into a Formula You Keep in the Workbook

The chat is useful once. The formula is useful every month. So the second half is asking Claude to hand the logic back as something that lives in Excel.

Ask it plainly: Write me an Excel formula for percent variance where column B is budget and column C is actual, that returns blank instead of an error when budget is zero, and that handles negative budget figures correctly. You get back something like:

=IFERROR((C2-B2)/ABS(B2),"")

The ABS around the denominator is the piece people miss. Without it, a budget of negative 5,000 against an actual of negative 4,000 reports as unfavorable when it was favorable — the sign flips and the column reads backwards on credits and contra accounts.

Then ask for the highlight rule, which is what saves you the scanning next month:

• Select your data range starting at row 2, for example $A$2:$F$200.
• Go to Home > Conditional Formatting > New Rule > Use a formula to determine which cells to format.
• Enter =AND($B2<>0,ABS(($C2-$B2)/ABS($B2))>=0.1) and pick a fill color.

Now every row more than ten percent off budget lights up the moment you paste next month’s export in. The gotcha is the dollar signs: the column letters are locked with $B and $C, but the row number stays unlocked at 2, and it must match the first row of the range you selected. Lock the row too and every cell gets tested against row 2 only — the rule appears to do nothing, or highlights everything, and it is almost always this.

Run the two together and the month-end hour becomes about ten minutes: paste, read the ranked list, edit the commentary, and let the highlighting tell you where to look next time.

Plenty more on the calendar

If you want to see this live, with a real workbook and your questions answered as we go, I am running Using Claude to Be More Productive in Excel on Thursday, October 22. We cover pasting ranges the way that keeps your structure intact, prompts that produce formulas you can trust, cleaning up a messy export, and building reusable rules so the work does not start over every month.

The full current schedule is on the front page of PCWebinars.com, and new dates go up as they are confirmed. Register for this one, or browse the others and find the session that fits what you are wrestling with right now.

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 the teaching, I build real databases and reporting systems for clients in Excel, Access, and Power BI, so the examples I bring to 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 was useful, there is a lot more where it came from — come see what is coming up at PCWebinars.com.