Why Standard Industry Codes Fail at Organizing Government Technology

Ask a procurement office to pull every technology contract a city has signed in the last five years and you'll get a report built on commodity codes. Ask whether that report tells you anything useful about the technology market — who the real vendors are, what category a system belongs to, how a permitting platform differs from a records management system — and the answer is almost always no.

That's not a data-entry problem. It's a classification problem. The two coding systems that govern how technology gets categorized in public procurement — NAICS and NIGP — were never designed to describe modern government technology. They were designed to count economic activity and to track spend. Used for anything beyond that, they collapse a sprawling, fast-moving market into a handful of buckets that tell you almost nothing.

Here's exactly where they break down, and why it matters if your job is to actually evaluate and buy technology.

NAICS: One Code for Nearly All Software

The North American Industry Classification System (NAICS) is the federal standard for classifying businesses by economic activity. It's a six-digit, hierarchical system — two digits for the sector, drilling down to a six-digit national industry. It runs the economic census, federal contracting set-asides, and most vendor registration systems.

It is also remarkably thin where government technology lives.

Software falls in Sector 51, "Information." Drill all the way down through the hierarchy to the most specific six-digit level, and the entire universe of packaged software collapses into a single code: 513210, Software Publishers. That one code is supposed to hold a 911 computer-aided dispatch platform, a parks-and-recreation reservation app, an enterprise resource planning system, a police records management system, a permitting and licensing platform, and a constituent-relationship tool. One code. No distinction between any of them.

The neighboring codes aren't any more granular. 518210 covers "Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services" — the entire cloud and hosting market in one line. Move over to Sector 54 and you get a handful of computer-services codes: 541511 (Custom Computer Programming Services), 541512 (Computer Systems Design Services), 541513 (Computer Facilities Management), and 541519 (Other Computer Related Services). That's effectively the full toolkit. Every technology company that sells to your government maps to one of maybe six codes.

To put that in perspective: NAICS has a dedicated six-digit code for Drive-In Motion Picture Theaters (512132), separate from regular motion picture theaters (512131). It distinguishes greeting card publishers from other publishers. But a body-worn camera platform, a water-utility SCADA system, and a court case-management system all share the same "Software Publishers" code.

NAICS isn't wrong — it's doing the job it was built for, which is measuring industries at the level of the national economy. It was simply never meant to tell a city IT director what's in a technology category or who competes in it.

If anything, NAICS is the modern option. Plenty of vendor registries and business-data products still run on its predecessor, the Standard Industrial Classification (SIC), which is even thinner and frozen in time. SIC was last revised in 1987 — before the commercial web — and was replaced by NAICS for federal statistical purposes in 1997. Its entire packaged-software universe lives under one code, 7372, "Prepackaged Software," a classification that predates SaaS, cloud, and mobile entirely. Wherever you still encounter SIC behind a vendor lookup, you're navigating government technology through a map drawn before most of it existed.

NIGP: More Detail for a Ballpoint Pen Than for an Enterprise Platform

If NAICS is the federal economic lens, the NIGP Commodity/Services Code is the one most local governments actually classify their purchases with. It's the standard taxonomy for classifying commodities and services across 33 states and thousands of local entities, and it's baked into most e-procurement and spend-analysis software.

NIGP is structured in tiers: a 3-digit class, a 5-digit class-item, a 7-digit class-item-group, and an 11-digit detail code. The full system holds roughly 274 three-digit classes (219 product classes, 55 service classes), about 8,700 five-digit item descriptions, and — at the most detailed level — over 250,000 line-item descriptions.

That sounds comprehensive. The problem is where the detail lives.

NIGP was born in the early 1980s out of a working group of state purchasing officials, and its DNA is physical goods. The granularity is staggering for commodities. The published example from NIGP's own documentation walks a single ballpoint pen down to code 620-80-21-035-4 — and the system distinguishes that fine-point, black-ink pen (035-4) from the otherwise identical blue-ink version (045-3), the green (065-1), and the red (075-0). Four codes, just to separate ink colors in one box of pens.

Now look at technology. Most government software and IT services land in a single 3-digit class — 920, "Data Processing, Computer Programming, and Software Services" — plus a few hardware-oriented classes. The 5- and 7-digit items underneath exist, but they're organized the way procurement pays for things (programming services, software maintenance, hosting) rather than the way the market is structured (the dozens of distinct application categories a modern city actually buys). The result is that NIGP can tell you the ink color of a pen in finer detail than it can tell you whether a software contract is for public-safety dispatch or parks scheduling.

That's not a knock on the procurement officers who maintain it. NIGP does exactly what a spend-tracking taxonomy is supposed to do. It just wasn't designed to be a map of the technology market, and it falls apart the moment you try to use it as one.

Why This Gap Actually Costs You

This might read like an academic complaint about taxonomy. It isn't. The thinness of these codes creates real, expensive problems for the people buying technology.

You can't see your own spend clearly. When every software purchase rolls up to "Software Publishers" or NIGP class 920, leadership can't answer basic questions: How much are we spending on public-safety technology versus finance systems? Are we paying three different vendors for overlapping capabilities across departments? The codes flatten exactly the distinctions a CIO needs to manage a portfolio.

You can't find the market. Search a cooperative contract or a vendor registry by these codes and you get an undifferentiated list — every company that checked the "IT" box. A code-based search can't tell you which vendors actually operate in the category you're buying, because the category doesn't exist in the taxonomy. You're back to evaluating the four vendors you already knew about.

You can't benchmark. "What do comparable cities pay for this?" is unanswerable through commodity codes, because the code lumps your $40,000 scheduling app in with a $4 million ERP. There's no apples-to-apples comparison when the basket holds apples, anvils, and aircraft.

New categories have nowhere to go. Real-time crime centers, drone-as-first-responder programs, counter-UAS systems, AI dashcams — these are real, active municipal technology categories. None of them have a code. They all disappear into the same generic bucket, which means the fastest-moving parts of govtech are precisely the parts the standard codes can't see.

What Organizing Technology for Government Looks Like Instead

The fix isn't a better commodity code. It's a different organizing principle entirely.

Commodity codes organize by economic activity (NAICS) or by how procurement pays (NIGP). Neither organizes by what a government buyer is actually trying to do, which is solve a department's problem. A public-safety director evaluating real-time crime centers and a public-works director evaluating water-system monitoring don't need to know they both technically fall under "Software Publishers." They need a structured view of their category — the real vendors in it, what peer cities paid, and how implementations went.

That's why market intelligence built for government organizes by department function, not supplier type or economic sector. Marketplace.city maintains 350+ market landscapes across 40+ department-function areas. Each function drills into specific categories — Public Safety alone is broken into distinct markets like Body-Worn Cameras, CAD/AVL, AI Dashcams, Counter-UAS Systems, and Real-Time Crime Centers — and each of those is its own organized view of a specific market, the kind of distinction NAICS and NIGP can't make.

But the deeper difference is what gets classified. NAICS, SIC, and NIGP classify the company — one business, one code. That's the original sin: a vendor is tagged once, by what kind of establishment it is, and everything it sells inherits that single label. Marketplace.city classifies the product capability — what a given solution actually does for a department — so the taxonomy describes the technology, not the seller.

That one shift changes everything downstream:

• A single product can appear on multiple landscapes. A multi-functional platform — say, a records system used by both police and the courts, or a payments engine that serves utilities and permitting — shows up everywhere it delivers a capability, instead of being filed once and lost.
• Portfolio vendors span dozens of landscapes. A company like Tyler Technologies or Motorola Solutions sells across ERP, courts, permitting, public safety, and more. Under "Software Publishers" that breadth collapses into a single code. Mapped by capability, each of their products lands in the specific market it competes in — so you see exactly where a broad incumbent overlaps with a focused point solution.
• You see the real competitive set in a category. Because placement follows capability, a niche specialist that would never surface under a company-level code sits right next to the platform giants in the same landscape. That's the full field of who you should actually be talking to — not just the vendors big enough to recognize.

The difference is the difference between a taxonomy that counts technology and one that helps you buy it. Standard industry codes will always have a role in accounting and compliance reporting. They were just never built to answer the question every IT leader actually has: what does this market look like, and who should I be talking to?

Curious what your last few years of technology spend would look like organized by category instead of commodity code? We'll map a category you're currently evaluating to the real market — vendors, pricing, and peer references — in a 20-minute walkthrough. Schedule a walkthrough →