Data cleaning
Why your POS reports lie to you
Your point-of-sale system adds up exactly what it was given. The trouble is what it was given.
The Idrak team··6 min read
Every month, thousands of shop owners in Jordan open a sales report, look at the top ten products, and make decisions: what to reorder, what to discount, what to stop stocking. Most of those reports are wrong. Not because the POS is broken, but because the data going into it is messy in ways no report is designed to notice.
One product, many names
When three cashiers enter the same can as بيبسي 330, ببسي and PEPSI 330ML, your report shows three products with modest sales instead of one bestseller. The real top seller drops out of your top ten, and you reorder it late. In the sample files we've looked at, roughly one product row in four was a duplicate or a spelling variant of another.
- Alef variants: أ, إ, آ and ا written interchangeably
- Taa marbuta and haa: لبنة vs لبنه
- Yaa and alef maqsura: المراعي vs المراعى
- Arabic and English names for the same item
Fils that look like dinars
A price typed as 350 instead of 0.350 turns a can of soda into a 350-dinar item. One line like that can inflate a day's revenue a thousand times over — or, if the system rejects it silently, make a sale disappear. Because Jordanian prices have three decimals, this mistake is common and easy to miss.
Dates in three calendars
Arabic-Indic digits (١٥/٠٣/٢٠٢٦), Hijri dates (15 رمضان 1447) and American-style 3/15/26 can all appear in the same column. Spreadsheets treat many of them as text, so they quietly fall out of any month-by-month comparison.
What to do about it
You don't need to retype anything. Start by measuring the problem: how many duplicates, how many invalid prices, how many dates that can't be read. That's what Idrak's free Data Health Check does in about a minute. Then fix the data once, keep a record of every change, and let the same rules run on next month's export automatically.