Our projects
Messy data, one clean analysis
Exports from several disparate systems, each in its own format, cleaned, matched and turned into one analysis on top.
Four sources in, one analysis out. Nobody touches a cell.
weekly-revenue-analysis · on uploadDone
1Sources arrive in every format
Done- crm-export.csv4,800 rows
- billing.xlsx2,250 rows
- sales-tracker (Sheets)1,900 rows
- finance-adjustments.csv650 rows
Different column names, date formats and currencies. The run takes them as they come.
2Rules clean and match
Done9,600of 9,600 records
all records checked
- Names, dates and currencies made consistent
- 312 duplicates merged across sources
- !4 records that don’t match flagged
3The analysis builds itself
Done6of 6 sections ready
- Posted to Slack and emailwith the file attachedDone
- Written summaryplain-English recapDone
- Records that didn’t matchexceptions listDone
Analysis ready in under a minute. Posted to Slack and email.Example run · illustrative data
Example run on illustrative data, with screen time compressed. Names and figures are made up.
What goes in, what comes out.
What you give it
- Exports from each sourceCRM, billing, a shared tracker, a finance sheet
- Whatever shape they come inDifferent columns, date formats, and currencies
It runs on its own
Starts when someone uploads the files
What you get
- One clean, matched datasetDuplicates merged, formats made consistent
- The analysis on topTables, movers, monthly changes, and a written summary
- A list of what didn’t matchSo a person decides, not the code
How it runs
- 1Someone uploads the files, or the run picks them up from a folder.
- 2Rules standardise names, dates and currencies across every source.
- 3Records are matched and de-duplicated; anything ambiguous is set aside.
- 4The analysis is built on the clean data and shared to Slack and email.
Runs on
A simple upload page, plus Slack and email. No new software for your team to learn, and $0 recurring costs.
What it isn’t
- A data warehouse project
- A dashboard nobody opens