Data Analysis with WorkBuddy: Clean, Chart and Conclude

By the end of this article you'll be able to hand a spreadsheet to WorkBuddy and get back cleaning, statistics, charts and conclusions — and know how to verify the numbers it gives you.

1. How Far It Can Take an Analysis

Routine spreadsheet work all runs in WorkBuddy: merging multiple sheets, deduplicating, filling gaps, aggregating by dimension, calculating year-over-year and month-over-month change, building pivot tables, producing bar and line charts, and writing the conclusion sections. It suits analysis where "the question is clear and the data is in the table". It is not suited to making judgements you have no data to support.

2. Upload and Authorization

On desktop you can drag files into the conversation or reference a workbook in the workspace with @; in cloud working mode files are processed in a sandbox. For sheets containing personal information, customer lists or payroll data, anonymize them before uploading, or upload only the columns you need. After processing, confirm the original file hasn't been overwritten.

3. How to Frame the Request

Write the question as one verifiable sentence, not "analyze this for me". For example: Calculate monthly sales by region and the month-over-month change, find regions that declined two months running, and base the cause analysis only on fields in the table. Then add three things: definitions (does it include returns, which date field), output format (table or chart), and how many conclusions you want.

4. Cleaning and Statistics: Have It Report Problems First

Start with a health check: how many rows and columns, the missing-value rate and outliers per column, duplicate primary keys, whether date formats are consistent. Set cleaning rules only after you have that report — for instance, whether blanks count as zero or get dropped, and whether an unusually large amount is a data-entry error or a real large order. Ask it to write the cleaning rules as a list so you can review them.

5. Charts: Decide What They Need to Say First

Charts serve the conclusion. Line for trends, bar for comparisons, stacked or pie for composition, scatter only for correlation. State in the instruction that "one chart tells one thing", and have it label units, time range and definitions. The default palette is fine — no need to chase something flashy.

6. Conclusions and Report Export

Have it write conclusions in three parts — observation, evidence, recommendation — with each conclusion noting which sheet and field it came from. When you need to deliver, export to Excel or Word and the charts travel with the document. Keep the version for leadership to one page and put the detail in an appendix.

7. How to Verify the Numbers

• Spot-check two or three totals by recalculating them yourself with filters or a pivot table.
• Check the total row count against the source table to see if anything was silently dropped.
• Confirm which period is the denominator for YoY and MoM — don't count incomplete months.
• Check that the maximum in the chart matches the table.

8. Common Pitfalls

Merged cells throw dimensions out of alignment. Dates stored as text break monthly aggregation. Inconsistent headers across sheets cause rows to be mismatched. Different bases for percentages can invert a conclusion. When you hit these, have it explain how it handled them before deciding whether to accept the result. As of 2026-11, credit consumption and per-file size limits follow what the client actually shows.

If you want to walk through uploading and asking first, see the hands-on tutorials.