Master data quality check
Look at the master data itself before an ERP, WMS or e-commerce integration. Drop an export of your material list and we find and score duplicate keys, bad barcodes, Excel damage and inconsistent units, while the file never leaves your device.
The file is read only in your browser; it is not uploaded to any server.
Files separated by comma, semicolon, tab or pipe; the delimiter, header row and UTF-8 are detected automatically (Windows-1254 is read too). Quoted fields are supported. Up to 200,000 rows and 60 columns are checked.
01
How to use it
A
Export the master data from your ERP or material management system as CSV; take barcode and material number columns as text.
B
Drop the file in the box or paste it; the column mapping is guessed automatically, correct it if needed.
C
Review the score and findings, download the findings as CSV and pass them on to the teams who fix them in the source.
02
What is checked?
Completeness: empty required fields (material no, description, unit) and placeholders such as N/A or a dash; the fill rate of every column. Uniqueness: duplicate material numbers, numbers multiplied by leading zeros or case, the same GTIN assigned to different materials, and descriptions repeated across different numbers (spelling differences ignored).
Barcodes: length (8, 12, 13, 14), GS1 check digit, placeholders made of a single digit, Excel damage such as scientific notation (5.01E+12) or a trailing .0, lost leading zeros. Format: leading, trailing and double spaces, non-printable characters, damaged character encoding, rows whose column count does not match.
Consistency: unit synonyms (ADET, AD, PCS, EA), text in numeric fields and mixed decimal separators, negative or zero weights, net weight above gross weight, and weights and dimensions far from the median of their material group.
03
Why before the integration?
When ERP, warehouse, e-commerce or serialisation systems are connected, errors in the master data multiply: if the same GTIN sits on two materials an order matches the wrong product, a key with a trailing space cannot be found in the other system, a barcode damaged in Excel is rejected by a marketplace.
These errors should be fixed at the source and kept from coming back, rather than patched during the transfer. The check shows by number which fields are affected and how much, which makes the clean-up measurable.
04
How to read the score
For each of five dimensions we compute the share of rows without a warning or error finding in that dimension; the overall score is their average. Info-level findings do not affect the score. The score is for comparing two files, or one file before and after clean-up; it is not a standard or certificate.
Not every finding is a real error. Service items may have zero weight, two materials with the same description may be genuinely different variants, and in a very varied material group the outlier warning may be a false alarm. Judge the findings together with your business knowledge.
FAQ
- Is my file uploaded anywhere?
- No. The file is read in your browser and checked by code running in your browser; it is not sent to any server or stored. The data is gone when you close the page.
- Which formats and sizes can it read?
- CSV, TSV and TXT files separated by comma, semicolon, tab or pipe; the delimiter, header row and UTF-8 (including BOM) are detected automatically, and Windows-1254 is read too. Up to 200,000 rows and 60 columns are checked; the rest is cut off and the report says so.
- What should I watch out for when exporting from Excel?
- Format the barcode and material number columns as Text and save the file as CSV UTF-8. In General or Number format, 13-digit barcodes turn into scientific notation such as 5.01E+12 and leading zeros are removed; this cannot be undone after saving and has to be re-extracted from the source.
- How are outlying weights and dimensions found?
- Within a material group with at least eight positive values, values that deviate from the group median by a factor of 10 or more are flagged; if no group column is mapped the whole file counts as one group. If more than a fifth of a group deviates, the group is not considered uniform and is not flagged. Very varied groups can produce false alarms.
- If the check digit of a GTIN is correct, is the number real?
- No. The check only looks at the length and the GS1 mod 10 check digit; it does not confirm that the number is registered under a GS1 company prefix or belongs to a product. A random number passes the check digit about one time in ten.
- What if columns are mapped wrongly?
- You can change the mapping in the section above the results; the checks rerun immediately. Checks related to an unmapped column are skipped and noted as information in the findings.
Let's get your master data ready for integration
We build the data flow between your CRM and ERP systems; cleaning the master data is the first step. Let's talk about your data and your goal in a free discovery call.