Sell spare parts online

A spare-parts buyer is under pressure. A machine has stopped, production is standing, and they need the right part now. Machines also break outside office hours: the maintenance technician who cannot reach you at 19:00 on a Friday calls the next supplier. Everything about this scenario comes down to identification, which means helping someone who knows the machine but not your article number find exactly the right thing, fast.

Why this is harder than it sounds

The assumption that sinks most spare-parts projects is that the customer will think in your article logic. They will not. They have a machine with a type plate, a serial number, and possibly a photocopied drawing from the manual. Your article number 4711-A means nothing to them. If the only entry point into your shop is a search box expecting a part number, you have built a catalogue for people who already know the answer.

So the project is an identification project, and identification runs into a hard fact: the machine in the customer's hall is not the machine in your current catalogue. It left your factory years ago in a specific configuration, possibly with options, possibly modified in service since. Showing the parts list for the current model produces confidently wrong orders, and a wrong part delivered to a stopped machine costs the customer another day. That is worse than showing nothing at all.

Then there is the source data. Spare-parts data comes from engineering, from ERP or PLM bills of material, and it is structured but written for engineers. Designations are internal shorthand, there are no images, no descriptions, and no distinction between the parts a customer may order and the ones only a service engineer should touch. Turning that into something a maintenance technician can navigate is the bulk of the project.

Two more that catch people out. Superseded parts: 4711-A was replaced by 4711-B in 2019 and by 4711-C last year. If the successor chain is not modelled, buyers will order parts you no longer make. And availability. For a stopped machine, "in stock" and "available in three weeks" are different decisions, and a stock figure copied overnight will be wrong at exactly the moment it matters most.

If you sell through dealers, settle the channel question before launch. A shop that shows local dealers, their services and their stock alongside the parts is an argument for the channel. A shop that bypasses them without a word is a conflict you will spend a year managing.

The levers

Let the customer start from what they know

What good looks like: the entry point is a serial number, a type designation, or a code on the machine itself. A photo shows where to find the type plate. From that input the buyer reaches the parts list for their machine, not for the model family.

What it depends on: modelling machines and series as first-class objects in the catalogue rather than as text in a description, and being able to resolve a serial number to a configuration. Today the platform gives you the reference entities and associations to hold that structure; the guided identification screen on the storefront is planned, so for now the entry point is search over those structured values.

1. The catalog data model
2. Reference entities for machines, series and assemblies
3. Attributes and attribute groups
4. Assets and asset families for exploded drawings, manuals and type-plate photos
5. Measurements and units

Model the part-to-machine relationship, with quantities

What good looks like: an assembly knows it contains four of a given seal. A nested assembly is navigable in both directions, from the drawing to the part list and back. The buyer can see which machines a part fits, and which part supersedes it.

What it depends on: getting the bill of material out of engineering in a usable structure, and deciding what customers may see. Multi-level structures are the normal case in special-purpose machinery.

1. Families
2. Variants and family variants
3. Prepare your import file
4. Import products in bulk
5. Data quality in practice

Tell the truth about availability and price

What good looks like: the buyer sees their contract price and the current stock figure, with a delivery statement they can plan around. Where a part is not stocked, the page says so plainly.

What it depends on: reading stock and conditions from the system that maintains them, and monitoring that connection. This is the point at which the ERP integration decides whether the shop is trusted.

1. Integration patterns
2. Deciding your system of record
3. Connect a system
4. Contract prices for a customer
5. Check which price a buyer sees
6. Stock accuracy
7. Monitor your syncs

Take the order without the service desk in the middle

What good looks like: an urgent order placed at 19:00 is confirmed immediately, is in the ERP overnight, and reaches dispatch with the same priority a phoned-in emergency would have had.

What it depends on: a checkout that handles urgency (express shipping, a required-by date, an order reference the customer's own system understands) and a fulfilment path that treats a shop order as real.

1. The order lifecycle
2. Checkout configuration
3. Shipping methods and zones
4. Process an order
5. Partial deliveries and backorders
6. Returns and credit notes

A sensible order

Phase 1: one series, all the way through. Pick the machine series that generates the most identification calls today. Model it completely: machines, assemblies, parts with quantities, drawings, successor chains, prices, stock, order path. Narrow and complete is the whole strategy. This phase depends on getting one usable bill of material out of engineering, which is usually the longest lead time in the project and should start on day one.

Phase 2: extend series by series, driven by call volume. Do not extend alphabetically or by product-management preference. Count what your service desk is asked about and work down that list. Each series is cheaper than the last because the model and the import route already exist. This phase depends on Phase 1 having proved the identification path with real customers; extending a path nobody has validated multiplies a mistake.

Phase 3: the hard cases and the channel. Serial-number resolution to as-built configurations, custom and one-off machines, codes on the machine that link straight to its parts list, and the dealer view. These are the items that make the shop complete, and each of them is a project. They belong after you have proof that the straightforward cases work.

Common mistakes

• Assuming the customer knows the article number. If your only entry point is a part-number search, you have built for the wrong person.
• Publishing the engineering bill of material as-is. Internal designations, service-only parts and no images make a parts list that no maintenance technician can use.
• Showing parts for the current model to the owner of a ten-year-old machine. Confidently wrong is worse than "please contact us".
• No successor handling. Buyers will order discontinued numbers, and every one of those is a manual correction plus a delay.
• Copying stock overnight. For spare parts, the stock figure is the promise. A promise from yesterday is not one.
• Starting with the whole installed base. It never finishes. One series, completely, then the next.

How you know it is working

• Share of spare-parts orders placed without any contact with your service desk. The headline metric for this goal.
• Wrong-part returns, absolute count and as a share of shop orders, compared with the same figure for phoned orders. This is the metric that decides whether the shop earns trust; watch it weekly during Phase 1.
• Time from arrival on the site to order placed. Long times mean identification is failing even when the order eventually happens.
• Zero-result searches containing serial or type numbers. Each one is a machine you have not modelled yet, named by the customer.
• Orders placed outside your service hours. Direct evidence for the availability argument, and usually the number that convinces internal sceptics.
• Identification calls per week to the service desk. Categorise them by series before you start, so that Phase 2 has a priority list rather than an opinion.
• Share of shop orders containing a superseded part number. Should trend to zero as the successor chains are modelled.

Related goals

• Publish a digital product catalog, the catalogue work underneath all of this.
• Run a customer portal, the natural next step once parts ordering works.
• Cut manual work, where the internal payback lands.

What you're building

A parts catalogue that a maintenance technician can navigate from the machine they are standing in front of: machine and series as structured records, assemblies with quantities, successor chains, drawings and manuals attached, live stock and the customer's own price. Orders go straight to the ERP with the urgency the buyer chose. The engineering bill of material feeds the model through Integration Studio; the guided identification screen (serial number in, parts list out) is planned, so today the entry point is structured search.

flowchart LR
  ENG["Engineering BOM (ERP / PLM)"]
  ERP["Your ERP"]
  subgraph IS["Integration Studio"]
    IMP["Import profiles"]
    WF["Workflows"]
  end
  subgraph CS["Commerce Studio"]
    P["Products: families, variants"]
    REF["References: machines, series"]
    ASSOC["Associations: fits, supersedes"]
    AS["Assets: drawings, manuals"]
    PR["Prices"]
    INV["Inventories"]
    O["Orders"]
  end
  subgraph XS["Experience Studio"]
    SF["Storefront"]
  end
  ID["Identification UI (planned)"]:::planned
  ENG -- "assemblies, quantities" --> IMP
  IMP --> P
  IMP --> REF
  IMP --> ASSOC
  ERP -- "conditions, stock" --> WF
  WF --> PR
  WF --> INV
  SF -- "urgent order" --> O
  O -- "order" --> WF
  WF -- "order" --> ERP
  ID -.-> SF
  classDef planned stroke-dasharray: 5 5

What you need

• One usable bill of material for one machine series, from engineering, with quantities — Bills of material and assemblies
• Machines, series and assemblies modelled as reference entities, not as description text — Reference entities
• Families and variants for the parts themselves — Families
• Drawings, manuals and type-plate photos attached to products — Assets and asset families
• Contract prices readable from the ERP for the pilot customers — Contract prices for a customer
• Stock read from the warehouse system at a cadence you can defend — Stock and locations
• A checkout with express shipping and a required-by date — Checkout configuration
• A decision on what dealers see and get, written down — Sell through marketplaces

Build it

1. Count a month of identification calls by machine series and pick the series at the top — How buyers identify a spare part
2. Request that series' bill of material from engineering on day one — Model a bill of material
3. Decide the model: families for parts, reference entities for machines and assemblies, association types for "fits" and "supersedes" — Plan your data model
4. Create the reference entities for the series and its machines — References
5. Define the association types you need — Association types
6. Create one assembly by hand, with quantities, to prove the model — Create a product
7. Prepare the import file for the rest of the series and load it — Import products in bulk
8. Attach drawings, manuals and photos to the assemblies — Assets
9. Model the successor chains for discontinued numbers — Compatibility and fits-with
10. Connect the ERP for contract prices and stock, and check one part for one customer line by line — Check which price a buyer sees
11. Configure express shipping and the required-by date in checkout — Shipping methods and zones
12. Run three pilot customers on the series and watch wrong-part returns weekly — The buyer's path: find a spare part
13. Extend to the next series from the call list, reusing the import route — Maintain bills of material
14. Put sync monitoring and a stock-drift check in place before widening — Monitor your syncs

Decisions you'll need to make

Related product areas

• Commerce Studio — Products · Families · References · Association types · Assets · Price lists · Stock levels · Orders · Shipping methods
• Experience Studio — Sites · Pages · Documents
• Integration Studio — Import/export · Credentials · Workflows · Runs