Clean a messy sales export and build a summary tab in Google Sheets

A writer gave Claude's Google Sheets add-on a 246-row sales export with five planted problems; it cleaned it to 240 orders, fixed #N/A lookups and built a summary that matched every checked total.

Done withClaudeinGoogle Sheets· Nov 12, 2026

What they did

Data by Toufik built a made-up Q3 sales export with 246 rows and five deliberate faults: six duplicate rows, dates as text in three formats, 44 misspelled regions, 44 prices stored as text and 16 #N/A lookups caused by trailing spaces. They worked out the right answers with a separate script first. Then, in the Claude for Google Workspace sidebar add-on running Opus 5.5 at medium effort, they sent four plain prompts in one chat: make a cleaned copy in a new tab, explain and fix the #N/A cells, add a summary tab with revenue by region and month plus a chart, and give a formula to check the total. The add-on works by running small scripts, each approved from a card in the default ask-first mode.

How it went

All four tasks came out right in one run. The cleaned tab had 240 rows and matched the answer file order by order on every field, after about three minutes and four scripts. Its first pass put every date a day early because of a timezone mismatch, which it caught with its own check and fixed. It traced the #N/A cells to trailing spaces and wrapped the lookup in TRIM, but only after asking, since it had been told to leave the original tab alone. The summary used live SUMIFS formulas and every total matched. The whole test took 11 approvals.

Worth knowing

This is one run on a practice file. The add-on is in beta on paid plans, works only on the open native Google Sheet (an .xlsx must be saved as a Sheet first), and one operation stops after six minutes. Claude said it had spot-checked only the first and last rows, so add your own check cells. Its edits are recorded under your name.

Try it

With Claude· 9 parts to fill in

This tab is a raw export of [what the data is]. Make a cleaned copy in a new tab called [name]: one row per [record], real dates, consistent [category column] values, numeric [amount columns], and a [derived column] calculated as [formula]. Do not change the original tab. Then tell me what you changed and how many rows you removed. After that, add a [summary tab name] with [totals by which dimensions] using formulas that point at the cleaned tab, and a cell that checks the totals tie back. Ask before every edit.

Square brackets are the parts to fill in. The prompt is ours; the story is theirs.

Where they said it

Related