Spare Parts Catalogues in B2B Shops: Assemblies, Not Search Boxes

A maintenance engineer is standing in front of a machine that has stopped, and does not need a bearing but exactly the bearing at item 9 of assembly 10.20 of a lathe with serial number 19-08812. A search box cannot answer that question, an assembly tree can. B2B internet trade by wholesalers and manufacturers in Germany has grown by 7 percent to 509 billion euros (ECC Cologne), and the spare parts business is one of the areas where the step from the catalogue PDF into the B2B shop has the clearest effect. Companies with digital spare parts management noticeably reduce their error rate and shorten their processes (Trendbook Smarter Service). This article shows how to build a spare parts catalogue that guides buyers through machines and assemblies instead of full-text search: what data it needs, what the tree looks like in the shop, which entry paths buyers actually use and in which order the work can be delivered.

Why the search box fails for spare parts

Full-text search assumes that the buyer knows the name of the product. With spare parts, that is exactly where the chain breaks: the maintenance engineer knows the machine, the assembly and the item number from the drawing, but not the description under which your system files the part. The same belt is called timing belt in the engineering bill of materials, belt, toothed in the supplier catalogue and drive belt in the maintenance manual. A search for one term does not find the other two, and a search for the generic term returns dozens of hits with no relation to the machine at all. Anyone who evaluates their own on-site search queries sees this pattern very quickly: many searches with zero or with very many hits, and a high share of sessions that end right afterwards.

The consequence is not a lost order but a fallback to the old channel. The customer picks up the phone, sends a photo by email or orders the wrong part on suspicion. Every one of these routes ties up inside sales, delays the repair and creates returns. Digital spare parts management addresses precisely this: companies that map the process end to end noticeably reduce their error rate and speed up handling (Trendbook Smarter Service). The lever here is not better relevance tuning of the search on its own, but a second, structured entry point alongside it.

The market and the margin behind the catalogue

B2B internet trade by wholesalers and manufacturers in Germany recently grew by 7 percent to 509 billion euros (ECC Cologne). The largest share of revenue is generated by the suppliers' own online shops, while the marketplace share did not grow any further in 2024 for the first time (ECC Cologne). The second finding of the same survey is the interesting one: the central hurdle is not demand but implementing an own shop (ECC Cologne). That is exactly where the spare parts catalogue sits, because it is the technically most demanding part of a B2B assortment and at the same time the one with the clearest retention effect. Anyone planning the digitalisation of sales should therefore not push it to the end of the roadmap.

The installed base that feeds this catalogue is large. German mechanical engineering employs around 1.3 million people, and 3.1 million work in the sector across the EU-27 (VDMA). The industry association represents around 3,500 predominantly mid-sized member companies in 35 trade associations (VDMA). European mechanical engineering generated around 871 billion euros in revenue in 2024, of which 293 billion euros were sold outside the EU single market (VDMA). Every one of those exported machines stays in service for decades and creates spare parts demand in a time zone where your inside sales team is not available. A catalogue that supplies the correct item number around the clock is therefore less a sales channel than a promise of availability.

Commercially, the spare parts and service business typically carries significantly higher margins than the new machine business (Trendbook Smarter Service). At the same time staff is becoming scarcer: at the end of 2024, around 5.5 million people worked in manufacturing establishments with 50 or more employees, 1.2 percent fewer than in the previous year (Federal Statistical Office). Mechanical engineering is the branch with the highest number of employees within manufacturing (Federal Statistical Office). When fewer people in spare parts sales handle the same volume of enquiries, self-service stops being a convenience feature and becomes a condition for the margin actually arriving.

How spare parts buyers actually start

Before you build a tree, it is worth looking at what information a buyer actually arrives with. In practice there are five recurring starting situations, and each of them needs its own entry point. They do not differ in their goal, but in what the user is holding at the moment the need arises.

• Machine type and serial number: the classic maintenance case. The user reads the rating plate and expects the shop to show only matching parts from then on. This is the path the assembly tree is built for.
• Exploded view and item number: the user has the drawing from the manual in front of them and is looking for item 9. They do not need names, they need a mapping from number to article.
• Part number: service management and procurement work from lists. What counts for them is quick entry of several numbers in one form, not navigation.
• Order history: wear parts are needed repeatedly. Someone who ordered the same timing belt eight months ago wants to reorder it from the history without walking the path again.
• Maintenance plan or wear part kit: where intervals are plannable, the object being sought is not a single part but a bundle. A predefined kit per service level saves collecting twelve items by hand.

All five paths have to end on the same article page. Technically that means one canonical URL per article, with the assembly path passed as context rather than as its own address. Otherwise a bearing installed in four machine series produces four competing pages with identical content. The rules are the same as with faceted navigation and its filter URLs: the article stays indexable, the navigation route to it does not.

The data basis: bill of materials, assembly, item

An assembly catalogue is ninety percent a data project and ten percent an interface. The good news is that the data usually already exists, just not in one place and not in a form a shop can read. Engineering, ERP and service documentation each hold a section of it. The catalogue needs the following six building blocks, and it needs them in full before the first page is rendered.

• Multi-level bill of materials: machine, assembly, sub-assembly, item. Two to four levels are usually enough. What matters is that the levels are named consistently and do not contradict each other between machine series.
• Item number as its own attribute: the item number is a property of the installation point, not of the article. The same timing belt can be item 7 in one assembly and item 3 in another.
• Validity range per installation point: from which serial number to which does this assignment apply? Without that range the catalogue shows parts that are not installed in the specific machine at all.
• Successor and replacement relations: design revisions happen. The discontinued number has to point to the successor, and the successor has to state from which revision it fits.
• Unit of measure and packaging unit: a sealing ring is supplied in packs of ten, two are needed. If the catalogue does not separate the two, wrong order quantities follow.
• Technical attributes: dimensions, material, hardness, tightening torque. They let the user confirm that they have landed in the right place in the tree, and they feed the filters later on.

This structure does not belong in the shop but upstream of it. A product information system as the leading source keeps the bill of materials, validity ranges and attributes together and hands them to the shop. Prices, availability and customer assignments, by contrast, come from the ERP system. Which fields run over which interface should be settled before the first line of code, otherwise someone ends up maintaining the assembly structure in the shop and the systems drift apart.

Assembly tree and exploded view in the shop

Once the model is in place, the interface is comparatively straightforward. The user selects machine type and serial number, receives the tree of assemblies and expands it down to the item. Next to it sits the exploded view with clickable item numbers. Both views show the same list and mark the same selection. It matters that the tree keeps its state in the address, so that a technician can send the path to a colleague without describing it.

A lean structure is enough for the handover between systems. What is decisive is that every item brings its assembly, its number, its quantity and its validity range with it, and that a successor appears where it is needed. An extract for a single sub-assembly can look like this:

{
  "machine_type": "DX-450",
  "assembly": "10.20 belt drive",
  "valid_from_serial": "19-04000",
  "items": [
    { "item": 7, "part_number": "44-0712", "description": "Timing belt", "quantity": 1 },
    { "item": 8, "part_number": "44-0813", "description": "Idler pulley", "quantity": 2 },
    { "item": 9, "part_number": "44-0914", "description": "Bearing block", "quantity": 2,
      "successor": "44-0915", "note": "from serial number 21-00500" }
  ]
}

Technically this structure can be maintained in Shopware 6 as a separate entity alongside the category tree. Categories continue to organise the sales assortment, the assembly tree represents the technical structure, and both point to the same articles. This separation matters because a marketing assortment follows different rules from a bill of materials: one is reordered when the campaign changes, the other stays stable for as long as the machine is in the field. How deep the customisation goes depends on the assortment; a Shopware agency should estimate the effort against the real data stock, not against the number of articles.

Serial number, machine record and order history

The real leap comes when the catalogue knows not only the machine type but the individual machine belonging to the customer. The catalogue then becomes a machine record: a list of the assets this customer operates, each with serial number, year of manufacture, design revision and order history. From then on the entry point is no longer a search but a selection from the customer's own inventory. For customers with several sites the record can be kept per site, so that the technician in building 3 sees only the assets they actually work on.

For that to hold, the catalogue has to know who is looking. Spare parts prices are rarely uniform in B2B: contract customers, dealers and end customers see different terms, and service flat rates depend on the maintenance agreement. The mechanics are the same as with customer-specific prices from the ERP system: the shop does not show a list, it shows the result valid for this customer. Large customers additionally require the route through their own procurement system; for that, the assembly catalogue can be provided as a punchout catalogue without creating a second set of data to maintain.

Implementation step by step

A spare parts catalogue rarely fails because of technology and frequently because of sequence. Starting with the interface means building a tree for data that does not exist yet. The following sequence has proven itself in practice and can be delivered in stages rather than in one large release.

1. Pick one machine series as a pilot. Not the largest, but the one with the cleanest bill of materials and an active service customer base. Twenty assemblies are enough to settle every modelling question.
2. Measure data quality before migrating. How many items have no validity range, how many discontinued numbers have no successor, how many articles have no image? These three figures determine the effort more precisely than any estimate.
3. Define the leading source. For every field exactly one system that may write. Everything else reads. Record this decision in writing, otherwise it will be taken again several times during the project.
4. Prepare the drawings. Raster images from old manuals are vectorised and given hit areas. This is the line item that delays projects, because it requires manual work.
5. Build the tree and the detail page. Interface work starts only now, but against real data. The detail page gets a visible validity note, successor and packaging unit.
6. Connect the machine record. Import the serial numbers of existing customers, define roles, link the history. Only now does the actual retention effect emerge.
7. Measure and refine. Searches without hits, drop-offs inside the tree and the share of orders placed from the history are the three metrics that show whether the catalogue holds up.

Alongside the catalogue work, it is worth looking at the adjacent processes. Mandatory product data requirements also apply in the spare parts area as soon as components with their own labelling are involved; the underlying logic resembles that of mandatory energy labels in the shop. And because spare parts orders are urgent in nearly every case, the order confirmation with a delivery date decides whether the customer calls or waits.

Establishing the catalogue as a service channel

A finished catalogue is not the end of a project but the start of a channel that wants maintaining. Every new design revision, every discontinued number and every machine delivered changes it. Four building blocks typically decide whether it is still accurate two years later.

The spare parts catalogue is therefore the point at which product data quality translates directly into service quality. Once the structure is clean, it can be reused for maintenance kits, retrofit offers and service contracts. If you would like to check how robust your bills of materials are for an assembly catalogue, and where the build can sensibly be cut into stages, talk to us: XICTRON supports manufacturers and wholesalers in building B2B assortments and connecting them to their existing systems. A short message is enough for a first assessment.