Every procurement team eventually hits the same wall: their spend data sits in an ERP with supplier names, dollar amounts, and GL codes — but no coherent category structure that actually supports analysis. The question of how to categorize that spend is deceptively hard, and the choice between adopting a standard taxonomy like UNSPSC versus building something custom has real downstream consequences for benchmarking, reporting, and sourcing decisions.
We've ingested spend data from a range of mid-market organizations while building Proculr, and the category structure (or lack of one) is consistently the messiest part of every data pipeline. Here's what we've learned about the tradeoffs.
What UNSPSC Actually Is — and What It's Good For
The United Nations Standard Products and Services Code is a hierarchical taxonomy with four levels: segment, family, class, and commodity. It covers roughly 55,000 commodity codes and is maintained by GS1 US. The structure runs from broad to specific — Segment 43 covers Information Technology Broadcasting and Telecommunications; drill down far enough and you reach specific codes for things like enterprise resource planning software or wireless access points.
UNSPSC is genuinely good at a few things. It's universally legible — any procurement professional can look at a UNSPSC code and understand roughly what category they're looking at. It enables cross-organization benchmarking because you're comparing apples to apples when you're both using the same code for the same commodity. It also integrates cleanly with e-procurement platforms: SAP Ariba uses UNSPSC natively, and many supplier catalogs come pre-coded.
For direct spend or any category where you're buying from standard catalogs, UNSPSC is close to a default choice. The taxonomy was built for this world.
Where UNSPSC Breaks Down
The problems start in indirect spend, and they compound as you move into tail spend categories.
Consider marketing services. A mid-market company with $80M in indirect spend might have $4M flowing through vendors they loosely call "marketing." UNSPSC code 80141600 covers "advertising" and 80161500 covers "public relations." But the actual spend includes a content agency, a freelance video production house, a demand-gen SaaS subscription, event sponsorships, and a PR firm. These land in wildly different UNSPSC families — some under professional services, some under software, some under events — and the result is that your "marketing" category is fragmented across the spend cube in a way that makes category management nearly impossible.
This is the core tension: UNSPSC classifies by what something is, not by how procurement should think about it. A company spending $300K with a single marketing agency wants to see that spend in one bucket when they're building a negotiation strategy, not scattered across three segments because the agency provides "advertising" and "consulting" and "digital media."
The taxonomy also ages poorly in fast-moving categories. Cloud infrastructure, API integrations, consumption-based software pricing — the UNSPSC codes covering these areas haven't kept pace with how modern software is actually sold and consumed.
The Case for Custom Taxonomy
Custom taxonomies solve the UNSPSC fragmentation problem by organizing spend around how procurement actually manages categories rather than what the product or service technically is. A well-designed custom taxonomy groups spend by sourcing strategy: who are the suppliers in this category, what does the RFP process look like, what are the market dynamics, and who owns the supplier relationship internally?
The practical upside is significant. When every transaction with a marketing vendor rolls up to a "Marketing & Creative Services" category — regardless of whether it's a retainer, project fee, or software subscription — your category manager gets a complete picture for negotiations. You can benchmark total spend in that relationship, identify where you have leverage, and build a brief that reflects the actual commercial situation.
Custom taxonomies also let you carve out categories that are strategically important to your specific business. A company that runs fleet operations will want "Fleet Maintenance" as a top-level category. A company with heavy R&D spend needs "Laboratory Supplies" broken out with far more granularity than any general-purpose taxonomy provides. UNSPSC forces everyone into the same structure; a custom taxonomy can be calibrated to where the money actually goes.
The Maintenance Problem No One Talks About
Here's the part that makes us hesitant to recommend full custom taxonomies without caveats: they require ongoing governance, and most mid-market procurement teams don't have bandwidth for it.
A custom taxonomy is only as good as its consistent application. When a new vendor comes onboard, someone has to decide which category it belongs to. When spend patterns shift — say, the company starts using a new category of spend that didn't exist in the original design — someone has to decide where it fits or add a new node. When people leave and the person who built the taxonomy is no longer available to answer questions, consistency erodes.
We're not saying UNSPSC is always better — it isn't. But maintaining a custom taxonomy is real work, and teams that adopt one without a clear governance plan typically end up with the worst of both worlds: a taxonomy that started custom but has accumulated so many edge cases and exceptions that it's effectively incoherent.
UNSPSC, for all its fragmentation issues, at least has a clear answer to "where does this go?" You look it up. Custom taxonomies require judgment, and judgment without governance produces inconsistency over time.
A Practical Hybrid Approach
What we've seen work in practice is a hybrid: use UNSPSC at the segment level (eight broad categories) as an anchor, then build custom classification at the category and sub-category levels where your business has concentrated spend or specific sourcing needs.
In concrete terms, this looks like: everything UNSPSC Segment 43 (IT) gets classified at the top level by that code, but within IT you have a custom breakdown that reflects your actual portfolio — Cloud Infrastructure, SaaS Subscriptions, Hardware, IT Professional Services. You're not trying to get to the commodity level with UNSPSC; you're using it as an orientation framework while the custom structure handles the nuance that actually matters for procurement decisions.
This hybrid approach also has a practical advantage when you want to benchmark against external data. When the taxonomy at the top is standardized, you can compare your IT spend concentration against peer companies. The custom sub-categories below that level are your internal operational layer — they don't need to be externally comparable.
How Proculr Handles This
When spend data comes into Proculr from an ERP export — whether that's SAP Ariba, Oracle NetSuite, or Sage Intacct — we apply a two-pass classification process. The first pass maps transactions to UNSPSC segment and family level using supplier name and description signals. The second pass applies a configurable category overlay that customers define based on how their procurement function is organized.
The reason for two passes isn't complexity for its own sake: it's because benchmarking requires a standardized classification layer to work at all. If we want to tell you that your spend with facilities maintenance vendors is at the 72nd percentile for companies your size, we need to be comparing the same category across companies. That requires UNSPSC-anchored classification on our end. But your negotiation brief is built from the custom category view, because that's what reflects the actual commercial situation you're walking into.
One thing we've learned building this: the quality of classification matters more than which system you use. A consistently-applied custom taxonomy will produce better procurement decisions than an inconsistently-applied UNSPSC. The enemy of spend analysis isn't choosing the wrong taxonomy — it's letting the taxonomy drift through inadequate governance until it no longer reflects reality.
Choosing Your Approach
A few questions that actually determine which direction makes sense for your team:
Do you need to benchmark externally? If yes, you need at least a UNSPSC anchor at the segment or family level. Pure custom taxonomies make external benchmarking nearly impossible without translation layers.
How much indirect spend do you have, and is it concentrated? If 40% of your indirect spend flows through a handful of categories that don't map cleanly to UNSPSC families, custom classification at those levels will pay for itself in better negotiation outcomes.
Who owns taxonomy governance? If you don't have a clear owner who will make classification decisions when edge cases arise, lean toward UNSPSC. Imperfect but consistent beats well-designed but inconsistent.
What tools are you using? If your e-procurement platform is UNSPSC-native and your supplier catalogs come pre-coded, the switching cost of going fully custom is significant. Work with what your infrastructure already supports.
The spend cube only produces useful insight when the underlying classification reflects how you actually manage categories. The taxonomy choice is worth getting right once — and then worth maintaining diligently, regardless of which direction you go.
