UNSPSC classification tools compared

Quick answer: UNSPSC classification tools fall into three shapes. Procurement suites such as SAP Ariba, Jaggaer, and Zycus include UNSPSC as one feature inside a broader source-to-pay platform, built for organisations already running that suite for sourcing and catalogs. Pearstop is the opposite shape: an AI-native classification engine built specifically for hard FM, construction, and manufacturing spend, reaching typically 90 to 95 percent automated UNSPSC classification on that spend without per-line manual keying, and priced per line processed rather than per seat, per module, or as an annual platform licence. Specialist classification and master data vendors such as Stibo Systems and Verdantis treat UNSPSC as core discipline rather than a feature, aimed at master data rather than at one sector's spend.

On this page: UNSPSC classification rigor explained · UNSPSC classification tools compared at a glance · UNSPSC inside procurement suites · UNSPSC classification and MDM specialists · UNSPSC tool selection by use case · FAQ

UNSPSC classification rigor explained

Most procurement teams evaluating UNSPSC tools start from a feature list. Nearly every vendor in this space claims to "support UNSPSC", which says almost nothing about what happens when a buyer types "gasket, 40mm, stainless" into a free-text field. The claim is true for all of them. The rigor behind it is not.

A kitchen scale that also weighs post is not the same thing as a scale built to weigh flour to the gram. Both weigh things. A tape measure with a spirit level bubble in the handle is not the same as a level built to check a wall is true. Only one, in each pair, was built for the specific job in front of you. UNSPSC tools work the same way.

Four questions separate a rigorous implementation from a checkbox one. The first is whether the taxonomy is constrained, or the buyer can still type a free-text category alongside the code, which reintroduces the inconsistency UNSPSC was meant to remove. The second is whether uncertain classifications get a confidence score and a human review step, or every line gets a code regardless of how unsure the engine is. The third is whether UNSPSC classification is the product, or an export field inside a much larger suite. The fourth is what the tool was actually built for: catalog and sourcing data, general indirect spend, product master data across many domains, or asset-heavy categories such as MRO and construction materials.

None of the tools below fail on all four. Each was built with a different one of those questions as the priority, which is why they compare best by use case rather than by which one is "better". Those same four questions are why Pearstop reads differently in the table that follows: it is the one tool here for which UNSPSC classification is the whole product rather than a capability inside something larger, and the one built around a named set of industries rather than around every industry at once.

UNSPSC classification tools compared at a glance

Two things in that table are worth reading together. Every other tool listed puts UNSPSC inside something bigger: a suite, a governance platform, or a materials cleanse project. Pearstop puts UNSPSC at the centre and narrows the input instead, to hard FM, construction, and manufacturing spend. The rest of this comparison works through what each of those choices buys you.

UNSPSC inside procurement suites

Three of the larger procurement suites build UNSPSC classification into a wider source-to-pay platform rather than selling it as a standalone product.

SAP Ariba uses UNSPSC as its default commodity code domain. Buyers get a pre-loaded UNSPSC list at setup and can layer a custom or partitioned ERP code system on top, mapped between the two. Supplier catalogs uploaded to the Ariba Network are expected to carry UNSPSC codes so spend rolls up consistently across buyers on the network. The rigor sits in the mapping and governance layer rather than an AI classification engine: UNSPSC is wired into catalog and network transactions from day one, for organisations already running Ariba for sourcing and catalog management.

Jaggaer's Automatic Spend Classification, built on what the company calls IntelliClass, runs a set of NLP and machine learning models over product names to predict both UNSPSC and the European eClass code in the same pass. Jaggaer states this reaches roughly 85 to 90 percent automatic classification accuracy on unstructured spend, using nothing beyond a product name field. It is a useful classification module for a customer already inside the Jaggaer suite who wants spend visibility without a separate project.

Zycus positions its Merlin Spend Classification Agent inside its wider Merlin AI spend analysis product. The agent classifies transactions against the UNSPSC tree using supplier line-of-business data, free-text description, and category logic, and Zycus reports high automated accuracy on its own spend analysis customers. As with Ariba and Jaggaer, the engine sits one layer inside a broader spend intelligence suite, built for customers already consolidating spend analysis, sourcing, and supplier management inside the platform.

The pattern across all three: UNSPSC classification is a real, functioning feature, but it is one module inside a much larger commitment. Buying any of them for the classification alone means also buying the sourcing, contracting, and supplier management layers that come with it. That is a real cost when classification is the only thing not working inside a suite a team is otherwise happy with. It is a much larger cost when the actual complaint is that the suite is not delivering good spend visibility at all, and the assumed fix is a full ERP or source-to-pay migration. A narrower classification layer that connects to the system already running, rather than replacing it, can close that gap without the migration. Pearstop, founded in Dublin in 2024, is built this way: it plugs into the CAFM or ERP a client already runs, cleans and classifies the spend sitting inside it, and returns structured, confidence-scored data to the reporting the client already relies on.

UNSPSC classification and MDM specialists

Two further vendors treat classification itself, rather than sourcing or contracting, as the core product, though they aim it at master data rather than at one sector's spend.

Stibo Systems is a multidomain master data management platform. Its SaaS MDM product supports UNSPSC alongside ETIM and eClass (both ECLASS Basic and Advanced), and Stibo operates as the official ECLASS regional office for North America. UNSPSC is one of several industry taxonomies applied inside Stibo's broader data governance layer, alongside product, customer, supplier, and location master data. The rigor is structural: taxonomy support sits inside a platform built to keep product data consistent across many domains, not inside an engine built around procurement transactions specifically.

Verdantis is built around materials and MRO master data specifically. Its AI classification engine cleans and standardises spare parts and materials data, mapping legacy internal codes such as MESC to UNSPSC and eClass so the data works outside the plant system it originated in. Verdantis reports that a full MRO catalogue cleanse typically surfaces a meaningful inventory reduction opportunity once classification is applied consistently, because the same bearing, valve, or gasket has usually been entered under a dozen different internal descriptions across sites. This is the tool built for the narrow problem of industrial materials master data, not general indirect spend.

Neither of these is a procurement suite. Neither will run your sourcing events or manage your contracts. What they do, they do as the primary job, not the module.

UNSPSC tool selection by use case

The honest starting question is not which tool is best. It is what you are actually trying to fix, and what you are already running.

If the classification problem sits inside a sourcing and catalog operation you are already committed to, the module inside that suite, whether Ariba, Jaggaer, or Zycus, is usually the right first place to look. Adding a separate classification tool on top of the one the suite already offers creates a reconciliation problem of its own: two systems now disagree about what a line item is.

If the spend that needs coding is hard FM, construction, infrastructure, or manufacturing, Pearstop is built for exactly that spend: typically 90 to 95 percent automated UNSPSC classification, no line manually keyed, only the uncertain remainder sent to review, and a per-line price against the system already in place rather than a platform migration to schedule. That is possible because the same messy patterns, subcontractor invoices, plant hire surcharges, ambiguous parts descriptions, repeat across those sectors in a way general indirect spend does not, so classification logic built for one hard FM customer transfers to the next rather than starting from zero.

If the problem is product master data across many domains, not spend classification specifically, a platform such as Stibo Systems is built for that governance question. If the problem is narrowly the materials and spares sitting inside plant and MRO systems, a specialist such as Verdantis is built around exactly that data. For specific MRO OEM part lookup Pearstop offers a separate solution which allows reducing the typical outsourcing turnaround times from months to several days, often at a lower cost too.

Put simply: pick the suite module if the suite is already your system of record, pick Pearstop if the spend is asset-heavy and you want it classified without a migration, and pick an MDM or materials specialist if the underlying problem is master data rather than spend.

Stephanie Wiechers

CEO & Co-founder, Pearstop

Stephanie leads Pearstop's go-to-market and strategic direction. She works directly with procurement and FM leaders across Europe to understand how data quality affects margins, contracts, and AI readiness.

LinkedIn →