Press Enter to search  ·  Esc to close

Still rebuilding the same report by hand every month?

Most corporate Excel reports are still built manually. Every month someone opens last month's file, deletes the old numbers, pastes new data, and adjusts everything by hand. It works until it breaks — wrong paste, overwritten formula, new rows that destroy the totals, or the original creator leaves the company.

The solution is to replace the manual steps with formulas that pull data automatically. Claude can't rebuild the entire report in one click, but it's excellent at analysing your process and writing the exact formulas to replace each manual step.

Manual report — fragile
Jan_Report
Feb_Report
Mar_Report ✕
1
Region
Revenue
2
North
£24,500 ← hardcoded
3
South
£18,200 ← hardcoded
4
West
£31,750 ← hardcoded
Dynamic model — self-updating
Raw_Data
Summary ✓
1
Region
Revenue
2
North
=SUMIFS(Raw!E:E,Raw!A:A,"North")
3
South
=SUMIFS(Raw!E:E,Raw!A:A,"South")
4
West
=SUMIFS(Raw!E:E,Raw!A:A,"West")

What Claude does — maps each manual step to a formula:

Manual step (what you do now)
Formula replacement
Copy revenue by region from raw export
=SUMIFS(Raw!E:E,Raw!A:A,A2)
Delete rows where Status = "Closed"
=FILTER(Raw!A:E,Raw!D:D<>"Closed")
Look up manager name from staff table
=XLOOKUP(B2,Staff!A:A,Staff!C:C)
Count active accounts per region
=COUNTIFS(Raw!A:A,A2,Raw!D:D,"Active")
  • You describe the manual steps and Claude identifies which can be replaced with formulas
  • Claude writes SUMIFS, XLOOKUP, FILTER, or dynamic array formulas to replace each step
  • Claude redesigns the structure so source data flows into the output automatically
  • Claude documents the new model so anyone on the team can maintain it
Not everything should be automated. Claude will flag steps that still need human judgment — unusual exceptions, manual sign-offs, decisions that depend on context formulas can't read. That's the right call.

Six steps — from a manual copy-paste process to a formula-driven model that updates itself.

Monthly_Report.xlsx
Raw_Data ✓
1
Date
Region
Product
Status
Revenue
2
01/06/2026
North
Widget Pro
Active
4250
3
02/06/2026
South
Widget Pro
Active
3100
4
03/06/2026
West
Widget Lite
Closed
1800
One row per transaction · no merged cells · consistent column names · real dates · numbers as numbers
01

Document your current manual process

Write down every step you currently do — be specific. "Copy column D from the export into column F", "Delete rows where Status = Closed", "Add subtotal by region". This list is your starting point for Claude.

02

Show Claude the structure

Describe your source data — columns, location, update frequency — and the final report — what it shows and who uses it. Paste small samples of both if possible. The more specific, the better the formulas.

03

Let Claude analyse what can be automated

Claude will tell you which manual steps can be replaced with formulas and which ones still need human judgment. Review the analysis before building anything — this shapes the whole approach.

04

Build the model one section at a time

Start with the most important parts — subtotals or key lookups. Ask Claude for the formulas, test them with real data, then move to the next section. This prevents mistakes from compounding across the model.

05

Standardise your source data

Dynamic reports need clean, consistent source data. Ask Claude to recommend the best table structure, column names, and formatting rules — then apply them to your raw export every month before pasting.

06

Create proper documentation

Once everything works, ask Claude to write a clear maintenance guide: what to paste each month and where, what the main formulas do, and what to check if something looks wrong. Save it inside the workbook.

Tip: Test the completed model with at least two months of real historical data before using it live. This catches edge cases — months with extra regions, missing data, or unusual values — that single-month testing misses.

Copy these prompts into Claude. Replace the bracketed placeholders with your actual process and structure.

Analysis prompt

Prompt — paste into Claude
I have a monthly Excel report I currently build manually. Here is the exact process: [describe each manual step] Source data has these columns: [list them] The final report shows: [describe output] Review this process and tell me: (1) which steps can be replaced with formulas, (2) the best formula approach for each, and (3) what changes I need in the source data structure.

Formula replacement prompt

Prompt — paste into Claude
I want to replace this manual step with a formula: "[describe the step — e.g. I copy total revenue by region from the raw data into a summary table]" Source data is on sheet "Raw_Data", columns A to F, starting row 2. Output should go to sheet "Summary", cell C5. Write the best formula for this.

Documentation prompt

Prompt — paste into Claude
The report is now formula-based. Write a clear maintenance guide that explains: (1) what to paste each month and where (2) what the main formulas do (3) what to check if something looks wrong
Checklist
Every manual step documented in plain English
Source data and output structure shared with Claude
Claude analysis reviewed — confirmed which steps can be automated
Formulas built and tested one section at a time
Source data structure standardised
Full model tested with at least two months of real data
Maintenance guide written and saved inside the workbook

Dynamic arrays in Excel 365: if you're on Excel 365, Claude will likely suggest FILTER, UNIQUE, or SORT instead of manual SUMIFS for some steps. These functions return multiple results automatically and eliminate entire categories of manual work. Tell Claude your Excel version and it will choose the right approach.