Published 18 August 2026. Rules and payer policies change. Claims here about a specific payer policy, a regulation or a statistic link to their source, so you can check whether it still says the same thing.

Part one of three on the fundamentals: who is who · the four processes · the terms that affect the work

Who is who in credentialing

Before any of the paperwork makes sense, it helps to be clear about who the parties are and what runs between them. There are five, and the relationships between them are not what a newcomer would guess: the person who provides the care is usually not the one who gets paid, and the patient is often not the subscriber.

An invented example runs through this series, because the words are easier to hold onto attached to somebody. None of these are real people or real businesses.

Maya Okonjo is a licensed professional counselor. She works at Cedar Street Counseling, a practice with six clinicians. Cedar Street does not do its own paperwork; it pays Northline Medical Billing to run credentialing and enrollment, which Northline also does for eleven other practices. That is one arrangement of several. Plenty of practices, groups and clinics run credentialing themselves, often with one person doing all of it, and everything below works the same way when they do. The work sits inside the organization instead of beside it. Tom Reyes sees Maya every other week, and his coverage comes through his wife's employer.

Nothing here is particular to behavioral health. Swap Maya, the counselor, for a physical therapist, a chiropractor, a podiatrist, an optometrist or a nurse practitioner, and every relationship below is the same. The specialty changes which payers matter and which rules apply on top; it does not change who contracts with whom.

The patient

The person in the session: Tom. They are at the top of the picture because everything below exists so that their appointment can happen and be paid for.

One distinction matters and is easy to miss. The patient is not always the subscriber. The subscriber, also called the policyholder, is the person in whose name the coverage is held. A patient may be covered as the subscriber's spouse or dependent child, and a child seen by a therapist is generally covered through a parent rather than in their own right.

This is not a fine point of vocabulary. The claim form asks separately who the patient is and who the insured is, and records the relationship between them — self, spouse or child. Getting that wrong is a common reason for a claim to come back.

The clinician

The person who provides the care: Maya. In behavioral health that is a therapist, counselor, psychologist or psychiatrist. In other practices it is a physical therapist, a chiropractor, an optometrist or a nurse practitioner. The job title changes; nothing else in this series does.

Every individual clinician has a national provider identifier, usually shortened to NPI, which is a ten-digit number that identifies them for the rest of their working life. There are two kinds, and the difference between them runs through everything else here.

"Type 1: For individual health care providers, such as physicians, nurse practitioners and sole proprietors. Individuals are only eligible for one NPI.
Type 2: For health care organizations, such as hospitals, nursing homes, and physician groups. Organizations can have multiple NPIs."

So a clinician has a Type 1. It belongs to them personally and follows them when they change jobs.

The practice

The business the clinician works for, and the one that gets paid: Cedar Street Counseling. If it bills in its own name as a separate legal organization — a corporation, an LLC, a group — it has a Type 2 NPI of its own, separate from every clinician working there, including Maya's.

That assumption is worth stating because it does not always hold. A clinician working as a sole proprietor is not a separate organization from the business, and generally uses their own Type 1 NPI rather than obtaining a Type 2 for it. The two-numbers picture below is the group-practice case, not a universal rule.

A person has exactly one NPI for life. An organization can have several, because a part of a large organization that bills on its own has to have its own number, so that whoever receives the claim can tell which part sent it. Those parts are called subparts; a psychiatric unit inside a general hospital is the usual example. So a hospital system does not have one NPI, it has many.

Smaller practices meet the same question. If Cedar Street adds physical therapy alongside the counseling, whether that creates a second number does not follow from it being a different kind of care. It follows from how Cedar Street decides to bill. Billing everything as one practice keeps one number; setting the physical therapy side up to send its own claims makes a second number compulsory rather than optional. The regulation leaves the decision to the organization but not the consequence, and the answer has to be written down, because it decides what goes on every claim from that point on. A claim carrying the wrong number is rejected, and the rejection will not explain itself in those terms.

Northline, which Cedar Street pays to handle all of this, has twelve practices to keep straight. It cannot work the answer out once and use it for all of them, because each practice has made its own decision about how it bills. Northline has to know, practice by practice, which numbers that practice holds and what each one is for.

One more point about the practice, because the rest of this series rests on it. The payer's agreement is generally with the practice, not with each clinician in it, and the payment goes to the practice. A clinician joining a practice that already participates does not receive a contract of their own at all — they are added to the one the practice already holds.

The payer

The insurance company or government program that pays for the care: Aetna, a Blue Cross Blue Shield plan, Medicare, a state Medicaid program.

Payers are not only the national names, and a practice that assumes they are will miss some of the ones that matter most to it. Health Plan of San Mateo is a managed care organization covering a single county in California, administering Medi-Cal — California's Medicaid program — for more than a hundred and thirty-seven thousand members there. For a practice in that county it is a payer like any other: its own enrollment, its own contract, its own dates. Being enrolled with a national insurer does nothing towards it.

Plan is nearly another word for payer, and people use it that way constantly, but the two are not the same thing. The payer is the organization. A plan is a particular coverage it sells, and one payer sells many of them — different employers, different benefits, sometimes different networks. That is why two people can both be covered by Aetna and still be working under different rules.

The tidy version does not survive contact with real names, though. Health Plan of San Mateo, mentioned just above, is an organization and not a product, and there are many like it. "Health plan" is used for the paying entity itself as often as "plan" is used for the coverage.

So there is no rule to learn here, only a habit worth having: when somebody says "the plan", work out from what they are talking about whether they mean the organization or the particular coverage. Most of the time it makes no difference. When it does, it is usually because the answer differs between two coverages sold by the same payer.

The network

Each payer keeps a set of practices and clinicians it has agreed to pay, under the agreements described above. That set is the payer's network. Aetna is not a network; Aetna has one. A clinician covered by one of those agreements is described as being in network for Aetna, and can be outside another payer's network at the same time.

This is a different use of the word from a hospital system such as Sutter Health or UCSF. Those are groups of providers under shared ownership, not a payer's list of who it will pay.

Whoever runs the credentialing

Somebody has to do this work, and that is the fifth part of the picture. Sometimes it is a person inside the practice or the group, with credentialing as their whole job or as part of a wider one. Sometimes the practice hands it outside, to a medical billing company, a credentialing service or a management company, which runs it for a number of client practices at once and charges for it. Northline, in the example, is the second kind.

The problem is the same either way and differs only in how far it spreads. Northline is keeping Cedar Street and eleven other practices straight, each with their own clinicians, payers and dates. A credentialer inside a single group is keeping straight every clinician, every payer and every date in that group. Both are holding more combinations than one person can carry in their head.

How a request actually travels

Naming the five parties does not say how anything moves between them, and the movement is where the difficulty lives.

Sooner or later, most applications need something that originates with the clinician. A current state license. A malpractice certificate showing the right dates. A board certificate. A curriculum vitae with no unexplained gaps in it. A signature or attestation on the application itself. Attestation in CAQH — the shared system where clinicians maintain and share their credentialing information, run by the organization that rebranded as DataSpring in June 2026 — generally every hundred and twenty days.

Whoever runs it does not originate any of it. Some may already be in its files or in CAQH; some will be out of date; and the missing piece is often still in the clinician's email, or in a drawer at home.

In this example, Northline does not go to Maya directly. It asks Cedar Street's practice administrator, who asks Maya, who is between appointments. The answer comes back the same way. Maya has no interest in credentialing and did not choose who runs it; from where she sits, it is a request for a document she is fairly sure she already sent last year.

So there are links in the chain — three of them when the work is done outside the practice, two when it is done inside — and each one is somebody with a different day job. That is why an application that takes twenty minutes to fill in can wait three weeks for the one certificate missing from it, and why "waiting on somebody else" is a state whoever runs credentialing needs to be able to see, age and count. Six items waiting on somebody else are not the same six if five of them are two days old and one has been sitting there for twenty-four. Nobody has done anything wrong. The request is simply sitting with somebody who has patients booked.

People are not the only thing being waited on. Payers do not ask for the same things in the same order, and they do not all take an application the same way. Some accept a form or a packet of documents, which can be assembled in advance from what is already known about the clinician and the practice. Some run the application inside their own portal, on their own sequence, and will close it if a message goes unanswered for long enough. Whoever runs credentialing has to know which kind each payer is, because the two need different work: one needs preparing, the other needs watching.

The short version

• The patient may not be the subscriber. A dependent is covered through somebody else's policy.
• The clinician has a Type 1 NPI. A practice that is a separate legal organization has a Type 2. A sole proprietor generally has only the Type 1.
• An organization can hold several numbers, one for each part of it that bills on its own. A person only ever has one.
• The agreement is usually with the practice. The clinician is checked separately and added to it.
• The documents come from the clinician, and when the work is done outside the practice the request usually reaches her through the practice rather than directly.
• A payer is an organization; a plan is a coverage it sells; a network is who it has agreed to pay. In speech the first two are used interchangeably.

Next: Credentialing, enrollment, contracting and privileging are not the same thing