data85
BlogJul 15, 20263 min readData Cleanup

Five passes every messy spreadsheet needs.

Spreadsheets get messy long before anyone decides to fix them. A second SKU column gets added to handle a one-off reconciliation. A new date format slips in for a single export. Currency symbols creep into half of one column. By the time the file is opened for a real decision, the totals don't match and nobody can say why. A structured, named cleanup is the move that pays for itself the first time the file is read.

← All posts
01

Pick a canonical key — and fold the variants back to it

A canonical key is the one row-identity you commit to — SKU plus date, invoice number plus line item, customer plus month. Every variant in the file gets folded back to that key before anything else happens. Two SKUs that look almost the same probably are the same product with a typo; two invoice numbers with different casing are the same invoice. Fold them once, here, and the rest of the cleanup is straightforward. Without this pass, totals silently double-count and downstream formulas lie about which customer shipped which order.

02

Standardise formats column by column

Pick a convention and apply it column by column. ISO dates (2026-07-12), USD with two decimal places, trimmed whitespace, sentence-case channel names. Don't accept 'we have a few formats' — accept one and rewrite. The file should look like one person wrote it on one day, even if it took a week and three exports. A consistent format is what makes the next formulas correct, and what makes the file readable in five minutes instead of fifty. The pass is mechanical; the payoff is permanent.

03

Reconcile signs and currencies with a stated rule

Clawbacks and fees need one sign convention — negative for money leaving, positive for money arriving — and the rule written down at the top of the sheet. Currency conversions need a stated FX rule ('all amounts in USD as of last business day of the month') so the next person can defend the number. Without these rules, totals silently cancel and reconciliation becomes theatre. With them, a reviewer can read one line at the top and trust every row below it.

04

Repair formulas and merge sheets last

Replace #REF!, ?, —, N/A, and TBD with placeholders that say what should go there. Repopulate totals so they tie to the source rows. Merge multi-tab files into one continuous range so quarter-to-date comparisons don't break at the sheet boundary. The pass that closes the file is the pass that lets the next person open it without a swarm of red flags. This is also where the cleanup gets reviewable — every cell that changed sits in row N, with a one-line note explaining the rule that was applied.

Related service

A clean file and a one-page changelog, returned in a single calendar week.

Data Cleanup

ready

A clean file and a one-page changelog, returned in a single calendar week.

No hire needed yet — one short email with a sample is enough to start.

3 min read
Flat scope · Fixed quoteOr send a sentence → same business day reply.