Autonomous data cleaning, normalisation and deduplication

How it runs, in three steps

1. 1The files come in as they are: exports from the webshop, the invoicing tool and the bank, plus five to ten supplier price lists, each one in a different shape.
2. 2The program repairs the broken accented letters and the three different phone number formats, recognises that “Kovács és Társa Kft.” and “Kovacs es Tarsa Kft” are the same partner, merges everything into one list, and compares the new prices with the old ones.
3. 3You get a tidy table, a plain answer to a question you typed in everyday words, and a Monday morning summary written in sentences; every number can be traced back to the file it came from.

Worth it when: It pays off first where the numbers sit in several systems at once, in webshops, wholesalers and anyone juggling many supplier price lists, and where somebody now spends half a day a week in spreadsheets. It only works if the data exists somewhere at all: from figures nobody ever recorded, nothing can be worked out.