How to Start a Group Therapy Practice: The Operational Playbook

One useful definition up front, because search results blur it: a group practice is a multi-clinician business — two or more clinicians operating under one organization with shared administration, scheduling, and billing. It is not the same thing as group therapy, which is a session format. This playbook is about building the business.

The transition from solo to group is a change of job, not a change of scale. As a solo clinician, your product is therapy. As a group-practice owner, your product is an operating system that lets other clinicians deliver therapy — intake that converts inquiries, schedules that fill, documentation that gets done, claims that get paid, and a margin that survives compensation. The owners who struggle are usually excellent clinicians who kept doing therapy full-time while the business ran itself. It does not run itself.

Expect the launch sequence to take three to six months from decision to first associate session. The pacing item is rarely the paperwork — entities and NPIs take days to weeks — it is payer credentialing, which is measured in months and cannot start until you have a hire and an entity. That is why the steps below front-load the business infrastructure.

Entity choice is state-specific in a way most articles undersell. Many states require licensed clinicians to practice through a professional entity — a PLLC or professional corporation — rather than a standard LLC, and several restrict who may own a practice that delivers clinical services at all. Some states also limit whether different license types (for example, an LMFT and a psychologist) can co-own one professional entity. Your two authoritative references are your state’s Secretary of State business filing office and your state licensing board; check both before paying a formation service, and have an attorney review ownership if anyone other than you will hold equity.

Once the entity exists, it needs its own identity separate from yours. That means an EIN from the IRS for taxes and payroll, a business bank account, and — critically for a healthcare business — a Type 2 (organizational) NPI from the NPPES system run by CMS. Your individual Type 1 NPI identifies you as a rendering clinician; the Type 2 NPI identifies the practice as the billing organization. Group payer contracts and group claims run on the Type 2. Registering it is free and takes days; do it as soon as the entity is formed.

Insurance changes shape too. Your individual malpractice policy does not automatically cover an entity employing other clinicians. Plan for an entity-level professional liability policy (with each clinician either covered under it or carrying their own, listed as required by contract), general liability if you hold physical space, and — once you have employees — workers’ compensation where your state requires it. Finally, every vendor that touches protected health information needs a business associate agreement with the practice; HHS publishes sample BAA provisions that show what those contracts must cover.

Published startup-cost data for group practices does not really exist, so treat any confident number — including ours — as a model. The table below shows modeled ranges for a two-to-four clinician launch in 2026, built from published software prices, typical service-provider quotes, and stated assumptions. The single most important line is the last one: working capital. Because payer credentialing and claim payment lag the launch by months, you should expect to fund three to four months of operating costs before insurance revenue arrives at full rate.

Compensation is the biggest recurring cost and the most common source of launch-stage legal trouble, so get the classification question right before the percentage question. W-2 employees can be scheduled, trained, and held to your documentation standards; the practice pays employer payroll taxes (7.65% for Social Security and Medicare) plus any benefits, and withholds income tax. 1099 contractors control their own methods and schedule — and the IRS decides which one your clinician actually is based on behavioral control, financial control, and the relationship, not based on what your contract says. A practice that sets a contractor’s hours, requires its EHR and note templates, and pays a revenue split looks a lot like an employer. Several states apply even stricter tests than the IRS. Get a written opinion from an accountant or employment attorney in your state before offering 1099 terms.

On the percentage: there is no authoritative published benchmark for group-practice splits, so treat the commonly discussed range as market color, not data. In practice-owner communities, clinician compensation typically lands between 40% and 65% of collected revenue — W-2 arrangements toward the lower end because the practice absorbs payroll taxes, benefits, and no-show risk; 1099 arrangements toward the higher end because the contractor absorbs them. Sanity-check any offer against the salaried alternative: the BLS median annual wage was $59,190 for substance abuse, behavioral disorder, and mental health counselors and $63,780 for marriage and family therapists in May 2024. An offer that models out below what a community agency pays, with less stability, will not survive its first renegotiation.

Worked example, with every assumption visible: assume $140 average collected per session (set yours from your actual payer contracts), a clinician completing 22 sessions per week, and 46 working weeks per year. Annual collections per clinician: 22 × $140 × 46 = $141,680. At a 55% split, the clinician earns $77,924 and the practice retains $63,756 — before employer taxes, software, rent, admin time, and unpaid claims. Run the same math at 18 sessions per week ($115,920 collected) and the practice’s retained share drops by about $14,000 per clinician. Compensation percentages get the attention; utilization does the damage. One assumption note: this launch-stage example uses $140 as an illustrative contracted rate, while the group practice profitability model on this site deliberately uses a more conservative $100 blended average collected across a mixed payer panel — the two figures differ by stated assumption, not by data, so set both from your own contracts.

A second clinician does not add revenue; a second clinician adds volume to four connected workflows — intake, scheduling, documentation, and billing — and revenue only appears if all four keep moving. Solo practitioners run this chain from memory. The whole point of building the system before hiring is that memory does not transfer.

Design each link with the same three questions: who owns it, what states can the work be in, and where do exceptions go? Intake: one owner moves every inquiry from new to scheduled, declined, or waitlisted, and no inquiry sits untouched past a set number of days. Scheduling: availability is maintained in one place, and a cancelled slot is refillable the moment it opens. Documentation: every completed session either has a signed note or appears on a needs-documentation list someone reviews weekly. Billing: every completed session becomes a claim or invoice within days, and denials land in a queue with an owner instead of a shared inbox.

Write this down as a one-to-two-page operating model before you evaluate software or interview candidates. It becomes your demo script for vendors and your training document for the first hire. Our group practice operations guide covers the full framework — ownership, states, handoffs, and management cadence — and the software evaluation guide turns it into a vendor test plan.

A launch-stage group practice needs seven systems, and the chain above should drive the choice: the practice management platform is the spine, and everything else attaches to it. The table shows the categories with modeled monthly costs for a two-to-five clinician group at July 2026 prices. One structural note from our software comparison: most practice management platforms price per clinician, so that line grows with every hire — a five-clinician group runs roughly $145–$395 per month on per-clinician platforms versus $59–$299 flat per practice on ClinicPro360’s tiers. Model the stack at the roster you plan to have in two years, not the one you have now.

One requirement applies across the stack: any vendor that creates, receives, maintains, or transmits protected health information — the practice management platform, telehealth, phone and fax, email if clinical content touches it — must sign a business associate agreement with the practice. Consumer-tier tools frequently will not. Confirm BAA availability before adopting anything, not after.

Hiring for a group practice runs on two clocks. The employment clock — write the role, recruit, interview, offer — takes four to eight weeks when it goes well. The credentialing clock — getting the new clinician onto your insurance panels under the group contract — commonly runs 60 to 150 days per payer from a clean application to an effective date, and it does not start until you have both the entity infrastructure and the signed clinician. Every week a credentialed-pending clinician sits idle is compensation without collections, so sequence everything around the slower clock.

The sequence that protects you: define the role and compensation model first (the section above), then recruit against a realistic promise — a ramped caseload, named supervision if applicable, and transparent split math. When you make the offer, immediately verify the license directly with your state board, collect the clinician’s Type 1 NPI, and get their CAQH profile current and re-attested, because payers pull from it and a stale profile is the most common self-inflicted delay. Submit payer applications the week the offer is signed, not the week they start.

Plan the gap deliberately. While panels process, a new clinician can see private-pay and EAP clients, take supervised or out-of-network cases where contracts and state rules allow, build intake familiarity, and complete training on your documentation and workflows. What they cannot do is bill your busiest panel — so match the start date, the ramp plan, and your working-capital line to the credentialing timeline you actually observe, payer by payer.

The launch-stage practice has one economic engine: completed sessions × collected rate × clinicians. Everything you monitor in the first 90 days is an early-warning indicator for one of those three factors. Define the terms precisely before you measure — booked sessions are not completed sessions, and billed is not collected — because the gaps between those words are where launches quietly fail.

The utilization math, worked: at $140 average collected per session and 46 working weeks, a clinician completing 18 sessions per week generates $115,920 per year in collections; at 22 sessions, $141,680; at 25 sessions, $161,000. The spread between a struggling schedule and a full one is roughly $45,000 per clinician per year under identical assumptions. For a three-clinician practice, moving average utilization from 18 to 25 completed sessions per week is worth about $135,000 in annual collections — more than most launch budgets in this guide. Utilization is not a vanity metric; at launch it is the business.

Watch five leading indicators weekly, each mapped to a workflow owner from your operating model: time from inquiry to first offered appointment (intake health — days, not weeks), completed sessions per clinician against their ramp target (utilization), no-show and late-cancel rate (schedule quality), notes signed within your documentation window (a leading indicator of billing delay), and claims submitted within days of session plus the denial rate when they return (revenue integrity). When one drifts, fix the workflow, not the person — at this scale nearly every miss is a systems miss.

The sequence matters more than the speed. This six-month plan assumes a solo clinician converting to a group with one to two hires; compress or stretch it to your market, but keep the dependencies — entity before NPI, systems before hire, credentialing before revenue — in order.