- Bruxel.AI Blog
- Source reconciliation: when your ERP and your spec sheet disagree
Source reconciliation: when your ERP and your spec sheet disagree
By Manuel Mattar4 min read
In short
Bruxel's source reconciliation keeps, next to the master value of each attribute, what each of your connected systems reports. The product page and a queue show you where they disagree, and you decide which one wins: by hand, in bulk or with a priority order per attribute. Reporting never changes the master on its own.
The problem it solves
A product's weight lives in your ERP, its dimensions in engineering's PLM and its spec sheet in a manufacturer's PDF. They almost never say exactly the same thing. The most common outcome is that whoever wrote last wins: an integration overwrites the field and nobody knows there was a difference, or where the value now in your store came from.
In PIM, ERP and DAM we explained which system answers which question. Source reconciliation solves something else: what to do when two systems answer the same question differently.
The idea: one master and one voice per system
Each attribute of a product in Bruxel has its master value, the one that goes into your downloads and your channels. In addition, Bruxel keeps what each of your connected systems says about that attribute, with the date it saw it. What a system reports does not touch the master, the history or completeness: it sits aside until someone decides.
That way you don't lose information. You see where they disagree and decide which one wins, without one system overwriting another.
How it works in Bruxel, step by step
1. Each system reports with its own key
A system reports its value through the public API, with a key that has permission to write attributes. The source is the system the key is linked to in Channels → Connected systems or, if it isn't linked to any, the key's name. The value is validated like the master: a measurement is converted to the attribute's unit, and sending it empty clears what that system said. The technical details are in the developers portal.
2. The product page shows where they disagree
Under each attribute that a system reported, the product page shows how many sources there are and whether they match the master, for example “2 sources · don't match”, with each system's value and when it was seen. “Use this one” moves that value to the master and is recorded in the history as “Reconciliation”.
3. The queue resolves in bulk
In Catalog → Source reconciliation you find every product whose systems disagree, with filters by family, attribute and source. You select the rows, choose “Use the value from” a system and click “Move to master”. Whatever couldn't be moved tells you why: that source didn't report the attribute, the attribute is protected, it is outside your access or a rule sets it.
4. Let one system win per attribute
If your organization has priority rules turned on, picking an attribute in the filter shows “Source priority”: the systems, from the one that wins most to the one that wins least. From the moment you save the order, every time a system reports, the highest one on the list that has reported writes the master on its own, and the history marks it as “Priority rule”. “Apply to existing” does the same with what was already reported. Saving the order changes nothing by itself, and the rule never writes a protected attribute.
5. You take it to your reports
In Reports, “Discrepancies between sources” downloads the full list, and the PDF quality report says how many are left to resolve.
Who it is for
For companies with more than one system that knows something about the product: an ERP, a PLM, a warehouse, a spec sheet someone uploads. A hypothetical example: an appliance maker sees that its ERP says a refrigerator weighs 85 kg and its PLM says 85.4 kg. With the queue filtered by the “weight” attribute, the Master data team decides that the PLM wins for measurements, saves that order and stops reviewing every new difference by hand.
If today you only have one source (your import file), this feature doesn't add much yet; start by completing your catalog.
What you need to get started
- Your systems registered in Connected systems, each with its own API key (creating keys is for admins).
- Attributes defined with their type and unit, so conversion and comparison work.
- Someone with the Editor role or higher to resolve, and an admin to set the priority order.
The source reconciliation guide covers each screen, and the reconciliation module page summarizes what it includes.
Honest limits
- The comparison is exact, after converting units. There is no tolerance: 285 and 285.4 don't match.
- It applies to text, number, measurement, date, yes/no and select attributes; not to multi-select ones.
- Bruxel doesn't go fetch the data from your systems: each system reports it through the public API.
- Picking a value changes the master in Bruxel, not in the source system; fixing it there is still that system's job.
- The queue is only visible to people without limited access to part of the catalog, and without a priority order everything stays a discrepancy until someone decides.
Frequently asked questions
Does what a system reports change the data that goes to my channels?
No. What a system reports is stored separately and does not change the master value, the history or completeness. The master only changes when someone picks a value or when there is a priority order for that attribute.
Does Bruxel treat 285 and 285.4 as equal?
No. The comparison is exact, after converting units: 285 and 285.4 do not match. A measurement in another unit is converted before comparing.
How does a system report its value?
Through the public API, with its own key and permission to write attributes. The source is the system that key is linked to in Connected systems or, if it isn't linked, the key's name.
Sources
About the author
Manuel Mattar
Founder of Bruxel.AI
In 2014 he founded Wolf Sellers, an e-commerce agency in Mexico City, and later Bruxel.AI, the AI-powered PIM.
See all posts by Manuel MattarTry it with your catalog
Create your free account and try Bruxel with your own products using the welcome credits.
Keep reading
· 4 min read
The catalog in Bruxel: attributes, families, categories and variants
How Bruxel organizes a catalog: typed attributes with rules, families that show what is missing, a category tree, the grid for editing, and variants.
· 4 min read
Importing and linking products in Bruxel: from file to catalog
How Bruxel imports your catalog from CSV or Excel, what it does with each column, how it builds related products, cross-sells and kits, and what views do.
· 4 min read
Versioned files: datasheets, certificates and packages in Bruxel
How Bruxel stores each product's documents: file types, versions with approval, virus scanning, expiry dates and ZIP packages with an index of every file.