How to Choose Parts Catalog Software as a Parts Distributor

Choose parts catalog software by testing it against the queries your team actually receives, not against a feature list. For a parts distributor, four criteria decide whether a tool gets used at all: how many brands it covers in one place, whether it narrows results by machine serial number, whether it shows the exploded diagram alongside the part number, and whether it returns a usable answer when the customer describes a part in plain language instead of naming it correctly. Two further criteria — onboarding effort and who on the team can actually operate it — decide whether it survives the first year. Everything else, from reporting to integrations to user management, matters only after those six hold up.

The three categories of solution

Most tools a distributor evaluates fall into one of three categories. They solve different problems and are frequently compared as if they were interchangeable.

OEM EPC (electronic parts catalog). The manufacturer's own parts catalog software — the closest thing to an authoritative source for that brand. An EPC typically contains the assembly tree, the exploded diagrams, the serial-number breakpoints and the supersession history for that manufacturer's machines, though how complete each of those layers is varies by model generation: older machines are usually documented more thinly than current production. Its limits are structural: one brand per system, access typically tied to a dealer agreement, an interface built around the manufacturer's assembly taxonomy, and search that expects you to already know the correct catalog term.

Catalog aggregator. A third-party platform that consolidates parts data from several manufacturers behind one interface. It solves the multi-brand problem for a distributor selling into mixed fleets. The trade-off is data provenance: coverage depth, update cadence and diagram availability vary by brand, and the further the data sits from the manufacturer's source, the more you have to verify what you sell.

AI assistant layer. A search and interpretation layer that sits on top of catalog data — from the manufacturer or aggregated. It does not replace the catalog; it changes how you interrogate it. The problem it targets is the gap between how a customer describes a part and how the catalog names it. Mayster is one of these: it searches Komatsu, Caterpillar and Volvo catalogs from a colloquial description and shows the positions it found together with the section drawing and the serial-number range, where the catalog contains them. When the description matches several variants, the assistant asks a follow-up question instead of guessing. It is built so that any part number shown comes from the catalog data rather than being written by the language model.

These categories are not mutually exclusive. Many distributors run an OEM EPC for their franchised brand and something broader for everything else in the yard.

Criterion 1: brand coverage against your actual mix

Pull the last few hundred parts enquiries and count the distinct machine brands. That distribution — not a vendor's brand list — is your coverage requirement.

Three questions cut through most coverage claims:

• Is coverage per brand complete, or partial? A brand present in the list may cover current production models only, with older machines missing. Older machines are often where parts distributors find margin: they are out of warranty and less often served by the franchised dealer.
• What is the model-level granularity? A catalog that resolves to a machine family but not the specific variant will produce plausible-but-wrong numbers on any machine with configuration options.
• How is the brand's data actually structured? Manufacturers do not organize their catalogs the same way, and the difference changes where you have to look. In the Komatsu catalogs, the engine is a separate catalog addressed by the engine's own number, so an engine-internal part is found there rather than under the machine. In the Caterpillar catalogs, engine parts sit inside the machine's own catalog, in sections such as ENGINE AR or CYLINDER HEAD GP. A tool that assumes one convention for every brand will send the user to the wrong place on the other one.

Test it directly: pick five older, less common machines you have quoted in the last year and search for a mid-assembly part on each. Coverage gaps surface immediately.

It is also worth calibrating your expectations of the underlying data. Manufacturer catalogs are not uniformly complete, and the gaps are not rare edge cases. In our own catalog data, a noticeable share of Komatsu positions carry no part number at all in the source, and our Caterpillar catalog grew close to twofold once it was pulled again from the source — which means the earlier version, sourced in good faith, had been missing a large part of the manufacturer's own catalog. Assume any tool you evaluate sits on data with holes, and judge it partly on how honestly it tells you when it has hit one.

Criterion 2: serial-number variant handling

This is the criterion that separates a catalog you can quote from a catalog you have to double-check.

For many construction machines — especially models built over a long production run — a single model has multiple variants of the same assembly, split by serial-number ranges. The same nominal component can carry several different part numbers depending on where the machine's serial falls. A tool that ignores this returns the first match and lets the customer discover the mistake on the workbench.

Construction machines do carry a standardised product identification number (the PIN, defined by ISO 10261), and the machine serial derived from it is what the catalog uses to split variants — so "the customer cannot give me a number" is rarely the real obstacle. The obstacle is whether the software does anything useful with the number once it has it.

What to look for:

• The tool asks for, or accepts, the machine serial number as an input — not just the model.
• Results display the applicable serial-number range next to each part number where the catalog carries one, so the person quoting can see which variant they are looking at.
• When the serial is unknown or ambiguous, the tool says so and shows the alternatives, rather than silently picking one.

Ask a vendor to demonstrate a part with known serial-split variants and watch what the interface does when you change the serial. That single test tells you more than a demo script.

Criterion 3: diagram availability

An exploded diagram is what turns a probable part number into a confirmed one. The customer points at the balloon on the drawing; the number follows.

Practical checks:

• Are diagrams present for the assemblies you actually quote, or only the popular ones? Coverage is never uniform, and not every catalog section has a drawing at all — some sections are parts lists with no illustration in the source. Ask what proportion of sections carry a drawing, and spot-check the assemblies you sell most.
• Are the callouts interactive? A static scanned image is usable. A drawing where clicking a numbered callout highlights the corresponding row in the parts table is faster and cuts transcription errors.
• Can you get from a search result to the drawing in one step? If the part number and the diagram live in different screens, people stop opening the diagram.

Criterion 4: how search handles colloquial input

Catalogs are indexed in the manufacturer's terminology, usually in English, usually as formal component names. Customers do not use those terms. They describe the part by function, by where it sits on the machine, or by a shop nickname. A catalog search that only does literal keyword matching will return nothing for a query that a parts specialist would answer in seconds.

This is the failure mode worth testing hardest, because it is the one that consumes counter staff time. Run these three query types against any candidate tool:

1. A functional description — the part named by what it does rather than what the catalog calls it.
2. A positional description — "the seal behind the cover on the left-hand side," with no component name at all.
3. A shop nickname — whatever your customers actually say in your local language.

A tool that handles the first and fails the second and third is matching keywords with a synonym list. A tool that handles all three is interpreting the query. The difference shows up in whether junior staff can use the system unaided.

Note also what the tool does when it is unsure. Returning "I am not certain, here is the assembly it belongs to — check the drawing" is a correct answer. Inventing a confident-looking part number is not, and a wrong number in a quote costs a return, a delay and credibility.

Criterion 5: onboarding effort

Onboarding effort is dominated by one question: does the system arrive with the catalog data already in it?

• Data already loaded — your work is limited to creating users and access. Ask the vendor for a reference timeline from a customer of your size, rather than accepting a number from a slide.
• You supply the data — you are now running a data project: sourcing catalogs, normalising formats, mapping brands and models, and keeping it current as manufacturers publish updates. Expect ongoing maintenance, not a one-off load.

The second model can be right if you hold proprietary data no vendor has. Just price it honestly, including the maintenance, not only the initial load.

Two smaller factors: whether it runs in a browser (no per-workstation installs, no VPN into a dealer system) and whether adding a user is something your own admin can do.

Criterion 6: who on the team will actually use it

A tool used only by the two most experienced parts specialists has not solved the problem — those two were already fast. The value appears when the rest of the team can answer without escalating.

Map the tool against your real roles:

• Counter and phone staff need speed and a low floor: describe it in plain language, get to the right section and a drawing.
• Field sales need it on a phone or tablet, on site, without a VPN.
• Workshop and service need the diagram and the serial-range check more than the price.
• Purchasing and back office need consistent part numbers flowing into the quote or the ERP.

If a category of user cannot get an answer without asking a colleague, the tool has moved the bottleneck rather than removed it.

Comparing the categories

Criterion	OEM EPC	Catalog aggregator	AI assistant layer
Brand coverage	One brand, closest to the source	Many brands, varying depth	Depends on underlying catalogs
Serial-number variants	Present in the source data	Depends on how data was ingested	Depends on underlying catalogs; check whether it filters by serial
Diagrams	Broadest for that brand; not every section carries a drawing	Varies by brand — ask for per-brand diagram coverage figures	Inherited from the catalogs it searches
Colloquial search	Depends on the product; test with your own colloquial queries	Depends on the product; test with your own colloquial queries	The capability it is built around
Onboarding	Dealer agreement, brand-specific training	Account setup	Account setup
Typical daily user	Experienced parts specialist	Parts department	Intended for the whole team — verify on your own test set before assuming non-specialists can work unaided

How to run the evaluation

Assemble a test set before you take any demo. Twenty to thirty real enquiries from your own records, written exactly as the customer phrased them, with the correct answer recorded from whatever you use today. Include the awkward ones: the vague descriptions, the machine with an unclear serial, the long-out-of-production model nobody stocks for.

Include a few queries that are just a number, in the formats your customers actually send. Part numbers are written differently by different manufacturers — a Komatsu-style number such as 600-185-3100, a Caterpillar-style one written as 1R-0658 or 1R0658 depending on who typed it, a Volvo-style one such as VOE17533661 — and a tool that only recognizes one of these shapes will fail on a query your team receives every week. These are examples of number formats, not a recommendation of specific parts.

Then run that same set against every candidate and score two things: how many produced the correct part number, and how many produced a confidently wrong one. The second number matters more. A tool that finds fewer parts but never invents one is safe to hand to a junior; a tool with a high hit rate and occasional fabrications requires an expert to verify every answer, which returns you to the bottleneck you were trying to remove.

Frequently asked questions

What is the difference between an OEM EPC and a catalog aggregator?

An OEM EPC is the manufacturer's own parts catalog software and is the closest thing to an authoritative source for that brand — it typically holds the assembly tree, the exploded diagrams, the serial-number breakpoints and the supersession history for that manufacturer's machines. Its limits are structural: one brand per system, access usually tied to a dealer agreement, and search that expects you to already know the correct catalog term. A catalog aggregator is a third-party platform that consolidates parts data from several manufacturers behind one interface, which solves the multi-brand problem for a distributor selling into mixed fleets. The trade-off with an aggregator is data provenance: coverage depth, update cadence and diagram availability vary by brand.

Does an AI assistant layer replace the parts catalog?

No. An AI assistant layer sits on top of catalog data — from the manufacturer or aggregated — and changes how you interrogate it rather than replacing it. The problem it targets is the gap between how a customer describes a part and how the catalog names it. Mayster is one of these: it searches Komatsu, Caterpillar and Volvo catalogs from a colloquial description and shows the positions it found together with the section drawing and the serial-number range, where the catalog contains them. It is built so that any part number shown comes from the catalog data rather than being written by the language model.

Why do engine parts sit in different places in Komatsu and Caterpillar catalogs?

Because manufacturers do not organize their catalogs the same way. In the Komatsu catalogs, the engine is a separate catalog addressed by the engine's own number, so an engine-internal part is found there rather than under the machine. In the Caterpillar catalogs, engine parts sit inside the machine's own catalog, in sections such as ENGINE AR or CYLINDER HEAD GP. A tool that assumes one convention for every brand will send the user to the wrong place on the other one, so it is worth testing an engine-internal part on both brands.

Why does one machine model have several part numbers for the same component?

For many construction machines — especially models built over a long production run — a single model has multiple variants of the same assembly, split by serial-number ranges. The same nominal component can therefore carry several different part numbers depending on where the machine's serial falls. Construction machines carry a standardised product identification number, and the machine serial derived from it is what the catalog uses to split those variants. The question to ask about any software is not whether the customer can supply a serial, but whether the tool does anything useful with it once it has it.

How many real enquiries should I prepare before booking a demo?

Assemble twenty to thirty real enquiries from your own records before you take any demo, written exactly as the customer phrased them, with the correct answer recorded from whatever you use today. Include the awkward ones: the vague descriptions, the machine with an unclear serial, and the long-out-of-production model nobody stocks for. Add a few queries that are just a number, in the formats your customers actually send — a Komatsu-style number such as 600-185-3100, a Caterpillar-style one written as 1R-0658 or 1R0658 depending on who typed it, and a Volvo-style one such as VOE17533661. These are examples of number formats, not a recommendation of specific parts. A tool that only recognizes one of these shapes will fail on a query your team receives every week.

Is the tool with the highest hit rate the safest choice?

Not necessarily. When you run your test set against each candidate, score two separate things: how many queries produced the correct part number, and how many produced a confidently wrong one. The second number matters more. A tool that finds fewer parts but never invents one is safe to hand to a junior member of staff, whereas a tool with a high hit rate and occasional fabrications requires an expert to verify every answer — which returns you to the bottleneck you were trying to remove.