What Is a CPT Code? A Founder-Friendly Explanation
A CPT code is a five-character identifier — almost always five digits, occasionally ending in a letter — that names one specific medical, surgical, or diagnostic service. The American Medical Association assigns and maintains the code set, and in the United States, a claim generally cannot be paid unless it references one.
Quick answer: CPT (Current Procedural Terminology) codes are the standardized labels US payers use to know what a clinician actually did. If your product's service doesn't map cleanly to an existing code, you don't automatically lose reimbursement — but you inherit a harder, slower, and less certain path to getting paid.
If you're building anything that touches a clinician's workflow — software that triggers a billable encounter, a device that generates monitoring data, a service delivered partly by a human and partly by an app — this three-letter acronym will show up in almost every conversation with a health system, a payer, or a reimbursement consultant. Knowing what it actually is, and what it isn't, saves you from two common founder mistakes: assuming a code will simply appear once your product is good enough, and assuming the absence of a code means the idea is dead.
How CPT codes work — from service to claim to payment
A CPT code is the "what happened" half of a medical claim; a payer combines it with a diagnosis code and a fee schedule to decide what, if anything, to pay. Understanding the mechanics behind that sentence is most of what a non-clinical founder needs.
Who creates and maintains CPT codes
The CPT Editorial Panel, convened by the American Medical Association, owns the code set. Physicians, medical specialty societies, device and drug manufacturers, and other stakeholders can formally propose a new code or a revision to an existing one. The panel reviews proposals, and accepted changes are published in an annual update, with a smaller set of interim updates issued during the year for urgent additions.
This matters for two reasons. First, CPT codes are not a government program — they're a licensed, copyrighted product of a professional association, which is also why you'll rarely see the full code book reproduced online for free. Second, getting a new code created is a real, multi-month-to-multi-year administrative process with its own evidence requirements — it is not something a product team can shortcut by shipping a good app.
The three-part code family: Category I, II, and III
CPT isn't one flat list. It splits into three categories with different purposes:
- Category I — the codes most people mean when they say "CPT code": established, widely performed procedures and services with a five-digit numeric format. These are the codes tied to standard payment.
- Category II — optional, alphanumeric codes used for performance measurement and quality tracking. They carry no direct payment value on their own.
- Category III — temporary, alphanumeric codes for new and emerging services, technologies, and procedures. They exist so payers and researchers can track how often something is used before deciding whether it earns a permanent Category I code.
A genuinely new kind of service — a novel remote-monitoring modality, for instance — will often live in Category III for a period of years while utilization and outcomes data accumulate, before a specialty society petitions for it to graduate to Category I.
From delivered service to paid claim
The billing chain, stripped to its essentials, runs in a fixed order:
- A clinician (or a service delivered under a clinician's supervision) performs something.
- A coder or biller translates that service into one or more CPT codes, paired with an ICD-10 diagnosis code explaining why the service was medically necessary.
- The claim, carrying those codes plus any modifiers, goes to the payer — a commercial insurer, Medicare, or Medicaid.
- The payer checks the codes against the patient's coverage, medical-necessity rules, and its fee schedule, then approves, partially pays, or denies the claim.
A code existing doesn't guarantee payment — coverage policy, documentation, and medical necessity all still apply — but a code not existing removes the standardized path entirely. Without one, a provider is left billing an "unlisted procedure" code, which typically triggers manual review, inconsistent payment, or an outright denial.
CPT vs. ICD vs. HCPCS: what each code actually describes
Founders new to healthcare frequently conflate these three code sets because they all show up on the same claim. They answer different questions, and mixing them up is a fast way to lose credibility in a payer conversation.
| Code set | Answers | Maintained by | Format | Typical example of what it covers |
|---|---|---|---|---|
| CPT | What service was performed | American Medical Association | 5 characters, mostly numeric | An office visit, a diagnostic test, a monitoring service |
| ICD-10-CM | Why the service was medically necessary | World Health Organization globally; adapted for US clinical use by the CDC's National Center for Health Statistics | 3–7 alphanumeric characters | The diagnosis or condition driving the encounter |
| HCPCS Level II | What supply, equipment, or non-physician service was involved | Centers for Medicare & Medicaid Services (CMS) | 1 letter + 4 digits | Durable medical equipment, ambulance transport, certain drugs |
In practice, CPT itself is also called "HCPCS Level I" — HCPCS is the umbrella system, CPT is its physician-service layer, and Level II fills the gaps CPT doesn't cover, mostly supplies and equipment. A typical claim carries at least one CPT code and one ICD-10 code together; the CPT code alone, with no diagnosis attached, tells a payer nothing about medical necessity and won't be paid.
A hypothetical example: billing a remote monitoring program
To make this concrete, walk through an illustrative (not real-world-reported) scenario: a startup builds a connected device that tracks a chronic condition and streams readings to a care team.
A program like this is typically not billed under a single code. Instead, it draws on a small family of related codes that separate the distinct pieces of work involved:
- One code for the initial setup and patient education on using the device.
- A separate code tied to supplying the device itself and collecting a minimum number of days of readings within a month.
- One or more additional codes for the clinical time spent each month reviewing the data and communicating with the patient about it.
Notice what this structure implies for a founder building in this space: reimbursement isn't a single yes/no gate on "the product," it's a claim built from multiple discrete, separately documented pieces of clinical work. A product that only automates the data collection step — and skips the clinician review and interaction the codes require — may find that the codes it hoped to bill under simply don't apply, because the underlying service the code describes never actually happened. The code describes a specific bundle of human clinical work; the software is a tool inside that bundle, not a replacement for it.
When CPT codes matter for startup validation
Treat "does an applicable CPT code exist for this service" as an early validation question, not a launch-week detail. It belongs in the same pass as your market-sizing and willingness-to-pay work, because it directly shapes your go-to-market timeline and who can plausibly pay you.
Three signals are worth checking before you invest heavily in a healthcare-facing product:
- An established Category I code exists for the service you deliver. This is the strongest position — you're billing into a known, priced pathway, though you still need to confirm the specific payers you're targeting actually cover and pay for it.
- A Category III code exists, but no Category I code yet. You can bill, but expect inconsistent coverage decisions across payers, and treat "getting to Category I" as a multi-year milestone dependent on utilization data you'll need to help generate.
- No code fits at all. This isn't automatically disqualifying, but it means your near-term revenue model probably can't lean on standard fee-for-service billing. Cash-pay, value-based contracts, or a B2B sale to a health system's operating budget (rather than its claims system) become far more realistic paths.
If reimbursement is central to your business model, this is exactly the kind of assumption worth stress-testing before you build — the same discipline covered in The Complete Guide to Startup Idea Validation in 2025, applied to a domain where a wrong assumption about billing can quietly invalidate an otherwise sound product. For the broader mechanics of how payers decide what to cover and pay in the first place, see Healthcare Reimbursement Explained for Startup Founders. And if you're trying to answer the narrower, more urgent question — "can I actually get my product reimbursed" — a structured way to work through that is laid out in How to Find Out If Your Digital Health Product Is Reimbursable.
Key Takeaways
- A CPT code names the service, not the diagnosis. It answers "what was done," while an ICD-10 code answers "why it was necessary" — a claim needs both to be payable.
- The American Medical Association owns the code set, and new codes go through a formal, multi-stakeholder petition process — not an automatic response to a good product.
- Three categories exist: Category I (established, priced services), Category II (optional quality-tracking codes with no payment value), and Category III (temporary codes for emerging services, used to build the evidence base for a future Category I code).
- HCPCS is the umbrella system; CPT is its Level I, and Level II fills in supplies, equipment, and services CPT doesn't describe.
- No matching code doesn't kill an idea, but it does rule out a standard fee-for-service path in the near term — plan around cash-pay, value-based, or enterprise contracts instead.
- Check code availability early, alongside market and pricing validation, not after the product is built.
Frequently Asked Questions
Who creates and maintains CPT codes?
The American Medical Association's CPT Editorial Panel creates and maintains the code set. Physicians, specialty societies, and manufacturers can formally propose new codes or revisions, which the panel reviews and publishes in annual (and some interim) updates.
What's the difference between a CPT code and an ICD code?
A CPT code describes the service a clinician performed; an ICD-10 code describes the diagnosis that made the service medically necessary. Claims almost always need both — the CPT code alone doesn't explain why the service was justified.
How does a new digital health service get its own CPT code?
It typically starts in Category III as a temporary, trackable code while utilization and outcomes data build up. A specialty society can then petition the CPT Editorial Panel to create a permanent Category I code once there's enough evidence the service is established and widely performed.
Can you bill for a service that has no CPT code at all?
Providers can submit an "unlisted procedure" code with supporting documentation, but payers typically review these manually, and payment is inconsistent and often lower or delayed compared with a standard, priced code. It's a fallback, not a reliable revenue path.