Solutions
Synopps software: one engine, three modules for master data tasks
Academy
Master data training built on real project practice
Synopps Data Health · platform module
Upload your material master extract and get a list of duplicates, descriptions brought to naming templates, and characteristics split into attributes. The system does not fix disputed cases on its own: it shows them to you.
Duplicate check3 pairs
Basis: part number and text match
Synopps Data Health is software for cleansing the historical records in your material master. It runs through an extract from your ERP, finds errors in descriptions, separates true duplicates from equivalents, assigns records to classes and brings descriptions to naming templates, extracting characteristics into attributes. Data comes in from Excel, CSV or directly from your system, and the result goes back. A person makes the call on every disputed case.
How it works
Every step is visible in the interface. The system proposes and shows its reasoning, and disputed cases go to review.
Excel, CSV or an extract from your ERP. Progress on the whole material master is visible in the project summary.
A set of rules runs over every description: unclosed brackets, mixed Latin and Cyrillic characters, double and trailing spaces, records with no meaningful content. Each check returns a status: no issues or needs review.
The system finds similar records and splits them into true duplicates and equivalents. They are different things and must not be merged.
Records are assigned to classes and given a naming template. Every proposed class comes with a confidence level, and low confidence goes to review.
Characteristics are extracted from the description and split into the template attributes: category and type, brand, model, size, standard.
The manufacturer and part number pair is checked against web search results. The record card shows which sites confirmed the item and what exactly they confirmed. After the check the result is exported back to your system.
Before and after
The examples show the kind of fixes the system makes. They are illustrations, not client data. The highlighted part is what the system found and fixed.
A Cyrillic letter inside a Latin word
Pump code 31 С EXCSA137CShighlighted: Cyrillic С
Pump code 31 C EXCSA137CShighlighted: Latin C
With a confidence level, low confidence goes to review
Pump KM 150-125-250/4-5 w/motor 18.5kW
Pumps and compressorsconfidence: high
Characteristics from the text go into the template fields
PLANER 500W 82MM M1902 MAKITA
Abbreviations in the description do not get in the way
Battery GP recharg. AA 2700mAh 270AAHC
Manufacturer and part number are checked against web search results
PLANER 500W 82MM M1902 MAKITA
Confirmedseveral independent sources
Review
The system never rewrites your master data silently. It proposes and shows its reasoning, the user confirms or corrects, and the decision is kept in the record history.
A correction, a duplicate pair, a class or attributes. Next to it, the reasoning: a code or text match, the class confidence.
Manufacturer and part number are checked against web search results. The record card shows which site confirmed what.
For example, "Elbow 45° D160" and "Elbow 90° D160": the system found them similar, and a person marked them as not a duplicate.
A new run does not overwrite an accepted result.
Duplicate decisionshistory
Decided by a person · kept in the record history
AI agent
The tool was built for work with agents from the start. An agent connects to it, goes through the same steps and makes decisions on disputed cases. Its actions are kept in the record history, and it cannot overturn a person's decision.
Who it is for
Cannot keep up with the backlog.
Duplicates make you buy what is already in the warehouse.
Before migrating to a new system, when you cannot carry the junk across.
You need to start the project with clean master data.
FAQ
This is a tool and you work in it yourself. The service is a project where we do the same work and answer for the result. The tool costs less, the service is faster and needs none of your time.
Excel, CSV or an extract from your ERP. A description and a record ID are required, and other fields are picked up if they are there.
They are not lost or overwritten. They are flagged for review and collected in a separate list that a person works through.
A duplicate is the same item created twice, and it gets merged. Equivalents are different items that can replace each other for some tasks, and they are linked. If you merge equivalents as duplicates, the material master gets shorter, and procurement starts buying the wrong thing.
It depends on the size of the material master and on how many records go to review. We give an estimate for your volume after we look at the extract.
Whichever suits you. The software can run inside your company's environment, so the master data never leaves your perimeter. Or we work on our side if you have no internal restrictions. This is agreed before the work starts.
Two ways
Upload the extract, go through the steps, decide the disputed cases and load the result back into your system.
Book a demoDo not want to do it yourself? We cleanse master data as a project, from audit to loading into your system. This is our master data cleansing service.
Go to the serviceWhat it looks like on real master data: our projects.
Request