Spare parts master data hygiene: speed lives in the data, not the server

Anyone who has run a parts business knows this scene. A mechanic searches for a front brake disc for a Golf VII 2.0 TDI. The system returns three hits of the same brand with the same part number, each under its own item code, two with stock and one on promotion. Which one is right? The salesperson calls the warehouse, the mechanic waits, the order slips.

This is not a server, design or search-engine problem. It is a master data problem.

In the automotive aftermarket, item data quality is often underestimated, even though it directly determines sales speed, stock accuracy and customer trust. This article explains where the mess comes from, what master data hygiene really means and how we solved it in TEC3.

Where the mess comes from

Spare-parts data enters a typical company from at least three directions that know nothing about each other.

The catalogue. TecDoc delivers standardised data: brand, part number, attributes, OE references and vehicle linkages. Excellent data, but it describes the item, not your stock or your price.

Suppliers. Every distributor sends its price list in its own format: CSV, XML, Excel, FTP or API. The same part has a part number without spaces at one supplier, with a hyphen at another and with a brand prefix at a third. The EAN is sometimes there, sometimes not.

The ERP. In the business system the item lives under an internal code created ten years ago, at the first purchase, often by hand. When the supplier changed, someone opened a new code and the old one stayed.

Import these three sources straight into a webshop and you get exactly what you entered: three records for one part. Every further import only deepens the mess.

What master data hygiene really means

Master data hygiene means that for every physical item there is exactly one record to which all stock, all price lists, all OE references and all vehicles are linked. One item, one truth.

To get there, items are not compared by eye but by rules, checked in this order:

1. EAN / GTIN, when present and valid, is the strongest key.
2. Brand plus normalised part number: before comparison we strip spaces, hyphens, dots and prefixes and unify letter case.
3. OE numbers and cross references from the catalogue confirm that it is the same part, even if part numbers differ by version.
4. Technical attributes (diameter, thickness, height, number of holes) are the final check: if the numbers match but the attributes do not, the records are not merged but flagged for review.

The last point matters. A bad merge is more dangerous than a duplicate: if the system merges two different parts, the mechanic receives the wrong item. That is why TEC3 logs every merge, keeps the source records and allows a rollback.

How TEC3 does it

At the core of TEC3 is the PIM matrix: a central item record with attributes, OE numbers, cross references, images and documents. All sources are mapped into it at import, not into the webshop.

Import runs in three steps. First normalisation: every incoming record gets a uniform part number, brand and unit format. Then matching by the rules above, with a confidence level for every match. Finally enrichment: the merged item receives stock per warehouse, price lists per supplier and customer discounts from the ERP.

The result is that the mechanic sees one Brembo disc 09.C881.11 in the catalogue, with all three stock locations, the best price for his discount group and the delivery time for each warehouse. He decides in one row, not three.

Because the matrix is separated from the sales channels, the same clean base feeds the B2B and B2C shop, exports to partner catalogues and the API. An error fixed once is fixed everywhere.

Why this is a question of speed

We often hear: "Our catalogue is slow, we need a bigger server." In practice most slowness is data-related.

If the system has to compare part numbers on the fly at every search, look up the EAN in three tables and work out which of the duplicate records has stock, it will be slow regardless of hardware. If matching is done at import, a search returns one record with stock and prices already prepared.

That is exactly why the TEC3 architecture is designed for catalogues of two million items with an average search response under one hundred milliseconds. Not because of bigger servers, but because the heavy lifting is done once, at import, not two thousand times a day at every click.

The same applies to the price exchange, smart reordering and price-class filters: all of these are only useful when they rest on one record per item.

What it means in practice

For the mechanic: one hit, real stock, the right price, no call to the warehouse.

For the dealer: less manual price-list editing, fewer wrong orders, fewer returns.

For the distributor: a clean export to partner catalogues and an API you can offer without embarrassment.

For management: stock in the webshop that matches the ERP and a margin calculated on the right item.

Seven questions before you choose a B2B spare-parts platform

1. How does the platform recognise that two records are the same item? By which keys and in what order?
2. What happens when part numbers match but attributes do not?
3. Is matching done at import or at every search?
4. Does it keep source records and allow a merge to be rolled back?
5. How are stock per warehouse and price lists per supplier mapped onto the merged item?
6. Does the same data base feed B2B, B2C and the API?
7. What is the search response at one or two million items, not on a demo base with twenty thousand records?

If a vendor has no clear answers, you will only move the mess from three systems into a fourth.

We built TEC3 so that we can answer them. If you want to see how it works on your data, write to us.

Frequently asked questions

What is spare-parts master data hygiene?

A state in which every physical item has exactly one record to which all stock, price lists, OE references and vehicles are linked. Duplicate records from the catalogue, suppliers and the ERP are merged by clear rules with an audit trail.

Which keys does TEC3 use to merge items?

EAN/GTIN first, then brand plus normalised part number, then OE numbers and cross references. Technical attributes are the final check: if they do not match, the records are flagged for review instead of merged.

Why is a clean item master linked to platform speed?

Because matching is done once at import, a search returns one record with stock and prices already prepared. That is how TEC3 keeps its response under one hundred milliseconds even with two million items.