Can the team map your actual services?
The operating plan should distinguish clinician types, therapy and psychiatry workflows, delivery settings, payer products, telehealth, integrated care, and any SUD or higher-acuity programs in scope.
Buyer and transition guide
Choose a mental health billing company by its operating model.
A mental health billing company should be evaluated as an operating partner, not a claim-transmission vendor. The decisive questions are who owns each handoff, how payer rules are controlled, how denials are classified, what the practice can see, and how open revenue is protected during transition.
- Scope before price
- Evidence before promises
- Ownership before escalation
Seven-point buyer checklist
What separates a billing vendor from an accountable partner.
A polished proposal can hide process gaps. Ask every potential partner to show how the work moves, where evidence is stored, who owns each exception, and what happens when the expected workflow breaks. These seven dimensions create a common comparison frame.
Who prevents the avoidable claim?
Define ownership for clinician enrollment, effective dates, benefits, authorization, referrals, patient estimates, and the communication path when readiness is incomplete.
Can every payer instruction be traced?
A production rule should carry its source, product, applicable setting, effective date, review date, owner, and approval status. A spreadsheet without those fields is not enough.
Does the team resolve and prevent?
Denials should be classified into actionable root-cause families, assigned to named work queues, reconciled to the balance, and returned to the upstream control that produced the failure.
Can leaders inspect the work?
The practice should be able to see held claims, authorization exceptions, rejected claims, denial inventory, A/R aging, next actions, source evidence, and unresolved dependencies.
Is access designed, not improvised?
Document systems, roles, least-privilege access, secure transfer methods, audit expectations, incident paths, subcontractor boundaries, retention, and the return or destruction of data.
Will decisions turn into assigned work?
Governance should produce decisions, owners, dates, escalations, and measurable follow-up. A meeting deck without the underlying queues and evidence does not create control.
Exhibit 1 · The scorecard
Score every bidder on the same seven dimensions, before you see a price.
Print this, run it once per bidder, and score from what they show you rather than from what they say. A dimension you cannot evidence scores zero, not two. The weights below are ASP-RCM's judgment about what actually predicts a working engagement in behavioral and mental health billing, and you should adjust them to your own practice before you use them.
0
Not present
No answer, or an answer that describes an intention rather than an existing process.
1
Asserted
Described in the proposal or on the call, with nothing shown that demonstrates it running.
2
Shown
Demonstrated live or in a de-identified artifact, but without a named owner or a standard attached.
3
Owned
Shown, with a named owner, a stated standard, and the record that proves the work happened.
↔ Swipe the scorecard
| Dimension | What a 3 looks like | Ask for this artifact | Weight | Score 0 to 3 |
|---|---|---|---|---|
| Specialty scopeDo they know your services? | The operating plan distinguishes therapy from psychiatry, telehealth from in-person, integrated care, and any intensive outpatient, partial hospitalization, substance use or ABA program you run. For ABA they separate BCBA and RBT authorization and billing rather than treating them as one pool. | A written scope map naming every service line you deliver, with the code family and authorization rule beside each. | 20 | ___ |
| Front-end ownershipWho prevents the claim? | Named ownership for clinician enrollment, effective dates, benefits, authorization, referrals, patient estimates, and the escalation path when readiness is incomplete before a session. | The escalation path in writing, including what happens when a session is booked against an incomplete authorization. | 15 | ___ |
| Rule controlCan an instruction be traced? | Every production payer rule carries its source, product, applicable setting, effective date, review date, owner and approval status. Rules expire and get re-verified rather than living forever. | Five rows from their live payer matrix, de-identified, showing all of those fields populated. | 10 | ___ |
| Denial disciplineDo they prevent or just rework? | Denials classify into actionable root-cause families, route to named queues, reconcile to the balance, and feed a documented change to the upstream control that produced the failure. | A de-identified denial work queue showing reason, age, owner, evidence, next action and escalation, plus one example of an upstream change they made. | 20 | ___ |
| VisibilityCan you inspect without asking? | You can see held claims, authorization exceptions, rejections, denial inventory, AR aging, next actions and unresolved dependencies yourself, on demand, without requesting a report. | A live screen share of the view you would have on day one, not a sample PDF of a monthly deck. | 15 | ___ |
| Data boundariesIs access designed? | Documented systems, roles, least-privilege access, secure transfer methods, audit expectations, incident paths, subcontractor boundaries, retention, and return or destruction of data at exit. | Their access model in writing, their subcontractor list, and the exit clause covering return and destruction. | 10 | ___ |
| Governance and transitionDo decisions become work? | Governance produces decisions, owners, dates and measurable follow-up. The transition plan inventories open revenue and assigns pre-cutover and post-cutover ownership for every work type. | A real governance pack from another engagement, de-identified, plus their written cutover and reconciliation plan. | 10 | ___ |
| Weighted total: multiply each score by its weight, then sum. Maximum 300. | 100 | ___ | ||
How to read the total, as a decision heuristic and not an industry benchmark: below roughly 150 the bidder is selling capacity rather than an operating model, and you would be supplying the control yourself. Between roughly 150 and 220 they are workable if you are prepared to own the dimensions they scored low on, and the contract should say so explicitly. Above roughly 220 the remaining question is commercial, not operational. Two bidders who score within about twenty points of each other are not meaningfully different, so decide on the dimensions that matter most to your practice rather than on the total.
Exhibit 2 · In-house, outsourced, hybrid
Outsourcing is the wrong answer more often than anyone selling it admits.
We are a billing company. We still lose more of these conversations than we win on purpose, because a practice that outsources for the wrong reason churns inside a year and is worse off than when it started. Read the third row of each column first: that is the row that decides.
↔ Swipe the table
| Question | Keep it in-house | Outsource it | Run a hybrid |
|---|---|---|---|
| What you are actually buying | Direct control of every step, and the ability to change a process the same afternoon you decide to. | A team that already knows the payer behaviour, and coverage that does not stop when one person resigns or takes leave. | Control of the front end where clinical context lives, and outside capacity on the back end where volume and payer knowledge dominate. |
| What it actually costs | Salary, benefits, software, clearinghouse fees, training, and the management attention nobody puts in the model. Also the single-point-of-failure risk when your one biller leaves. | A fee, plus the internal effort of feeding the relationship, which is always more than the proposal implies. Budget real time for the front-end handoffs. | Both, at smaller scale, plus the cost of a clean boundary. A hybrid with a fuzzy boundary is the most expensive of the three. |
| When this is the right answer | Your volume is small and stable, one person genuinely has capacity, your payer mix is narrow, and your denial rate is already low. Also when your clinical and administrative work are so entangled that no external team could separate them without a documentation project first. | Your denials cluster in payer behaviour rather than in your own documentation, your volume has outgrown one person, you are adding clinicians or locations, or your current billing knowledge sits entirely in one head. | You want to keep eligibility, authorization and the client financial conversation close to the clinical team, because those depend on clinical context, and move claims, denials, appeals and AR to a team with more payer reach. |
| When it is the wrong answer | Wrong when your biller is also your office manager, your scheduler and your intake coordinator, because billing is the task that silently loses. Wrong when nobody in the practice can tell you the denial mix from memory. | Wrong when the real problem is your clinical documentation, your enrollment status, or your fee schedule, because no billing company can fix those from outside. Wrong when you are outsourcing to avoid a decision rather than to execute one. Wrong when you will not staff the front-end handoffs the vendor depends on. | Wrong when the split is drawn by convenience rather than by process, because every unowned handoff becomes a denial that both parties blame the other for. |
| The failure mode to watch | Billing quietly falls behind during a busy clinical stretch and nobody notices until timely filing has run out on a batch. | The practice stops looking. Reports arrive, nobody reads them, and by the time someone does the AR is nine months old. | Both parties assume the other owns the exception. The denial reason that nobody owns is the one that repeats. |
| How to test the decision | Can you state your denial mix, your AR over 90 days, and your unbilled session count right now without opening anything? | Would the problem you are trying to solve still exist if your current biller had twice the time and twice the payer knowledge? If yes, outsourcing will not solve it either. | Can you write the boundary as a list of processes with a single named owner each, with no process appearing twice and none missing? |
If the honest answer for your practice is in-house, that is a legitimate outcome of this page and we would rather you reach it here than six months into an engagement. If you want a second read before you decide, send a de-identified process map and a senior partner will tell you which column you are in, including when it is the first one.
Proposal normalization
Compare every bidder on the same process map.
Different proposals bundle work differently. Normalize them before comparing price. Otherwise, one bidder may include enrollment, denial appeals, patient balances, and legacy A/R while another quietly excludes them.
Map scope
List every revenue-cycle process and state whether it is included, excluded, or shared.
Name owners
Assign the practice, billing company, technology vendor, payer, or another party.
Define evidence
Specify the queue, source, log, report, or record that proves the work occurred.
Set standards
Agree on aging, response, escalation, review, and reconciliation expectations.
Price the same scope
Compare commercial terms only after services, exclusions, and dependencies match.
Exhibit 3 · How the three pricing models behave
Price the model, not the number. The shapes diverge as you grow.
Three structures dominate: a percentage of collections, a fee per claim, and a dedicated staffing or full-time-equivalent arrangement. At one point on the volume curve they can quote to nearly the same monthly cost. They do not stay there. The chart shows how each behaves as volume grows, in cost per claim, which is the view that actually decides. There are no rates on either axis, because your rates are yours to negotiate and we do not publish figures we cannot source.
↔ Swipe the chart
↔ Swipe the table
| Model | What it aligns | What it quietly penalizes | The clause that protects you |
|---|---|---|---|
| Percentage of collectionsA share of what posts | The vendor only gets paid when you do, which aligns them with denial work, appeals and underpayment recovery rather than with claim throughput. | Your own improvement. Renegotiate a better fee schedule, add a higher-paying service line, or recover a large legacy balance, and the fee rises with it for work that did not get harder. | Define exactly what counts as collections. Exclude patient self-pay you collect at the desk, capitation, incentive payments, and legacy AR unless the vendor is working it. Set a step-down at defined volume or revenue thresholds. |
| Per claimA fee per submission | Predictability. Your cost tracks claim count, so budgeting is simple and improving your rates does not raise your billing cost. | Effort per claim. A resubmission, an appeal, and a clean first-pass claim can all count the same, so there is no economic pull toward the hard work of preventing the denial. | Define whether corrected claims, resubmissions and appeals are billable events. Attach a first-pass acceptance standard and a denial-resolution standard so throughput alone does not satisfy the contract. |
| Dedicated staffingNamed people, fixed cost | Capacity and continuity. You get identified people, and at sufficient volume the cost per claim can fall below both other models. | Volume below break-even, and volume above it. Below, you pay for capacity you are not using. Above, quality degrades until someone adds a person, and the addition is a renegotiation. | Name the people, state the coverage plan for leave and attrition, and define the volume band the staffing is sized for with an agreed process for what happens at each edge of it. |
The chart shows the shape of each model, not rates. No monthly cost, percentage, per-claim fee or salary figure is asserted anywhere on this page, and the crossover points on the chart are illustrative positions rather than measured thresholds. Where the curves actually cross for your practice depends on your negotiated terms, your average payment, your claim volume and your denial rate, and you can only find it by modelling all three against your own last twelve months.
What a serious discovery call covers
The partner should inspect causes, not just totals.
A useful discovery begins with the practice structure: legal entities, locations, clinicians, services, payer products, EHR, clearinghouse, payment channels, current vendors, and the boundary between administrative and clinical work.
Then it examines the current claim flow. How are benefits and authorizations captured? Which claims are held before submission? How are rejections, requests for information, denials, and underpayments classified? Who owns patient balances? Which legacy balances will transfer?
Finally, the parties agree on a verified baseline and measurement method. Financial outcomes should not be promised without understanding contracts, payer mix, starting backlogs, enrollment issues, documentation, service mix, and the exact scope of responsibility.
Ask to see a de-identified work queue.
A workflow sample should show reason, age, owner, evidence, next action, dependency, and escalation path. That reveals more than a list of software features.
ProcessScope question to settle before contracting
Enrollment
Who submits, follows, validates effective dates, and resolves claims affected by enrollment?
Benefits and PA
Who verifies, records limitations, obtains authorization, tracks reviews, and communicates exceptions?
Claims
Who owns edits, documentation dependencies, submission, rejection correction, and proof of timely filing?
Denials
Who researches, corrects, appeals, documents decisions, and changes the prevention rule?
Balances
Who posts, reconciles, works payer A/R, manages patient balances, and handles legacy inventory?
Transition discipline
Protect open revenue while the operating model changes.
A transition should inventory the revenue cycle before any system or responsibility changes. The inventory includes payer and clinician enrollment, service locations, authorizations, interfaces, clearinghouse and payer portal access, open claims, denial queues, unapplied cash, refunds, patient balances, and A/R by age and owner.
Each item needs a cutover rule. Define which party works pre-cutover claims, which party posts later remits, how appeal deadlines are protected, how patient communications remain consistent, and how totals will be reconciled. Keep a daily decision log during the cutover period.
GateEvidence required before moving forward
Inventory
Payers, clinicians, locations, access, interfaces, claims, denials, cash, and balances reconciled.
Ownership
Pre-cutover and post-cutover responsibility defined for every open work type.
Access
Approved roles, secure methods, test transactions, and revocation plan documented.
Parallel check
Representative claims and remits traced through the new workflow before full release.
Reconciliation
Control totals, exceptions, owners, and sign-off criteria documented for cutover.
Exhibit 4 · Red flags
Six things a bidder says, and what to ask instead.
None of these are proof of a bad partner. Each is a claim that cannot be evaluated as stated, and each has a replacement question that can be. Ask the replacement and score the answer, not the claim.
A guaranteed collection rate, denial rate, or revenue lift quoted before anyone has looked at your data. No result can be assumed across practices: it depends on your starting backlog, payer mix, contracts, enrollment status, documentation and the scope you actually assign.
Ask insteadWhat baseline would you measure me against, how would you establish it, and what would you commit to in writing only after you have seen ninety days of my remits?Named clients used as proof, especially with numbers attached. Either the results are not attributable to the vendor alone, or the reference practice had a different starting position than yours, or both.
Ask insteadGive me two references at my size, in my specialty, including one engagement that did not go well, and tell me what you changed afterward.A price quoted before the scope is mapped. This is how one bidder ends up including enrollment, appeals, patient balances and legacy AR while another quietly excludes all four, and both look comparable on the page.
Ask insteadMark every revenue-cycle process as included, excluded or shared, sign that page, and then quote against it.Software shown in place of process. A dashboard is not a control. The question is never whether the tool can display a denial queue, it is who works it, against what standard, and what happens when they do not.
Ask insteadShow me a de-identified work queue as it stands right now, with reason, age, owner, evidence, next action and escalation on every row.Vagueness about who does the work and where. Not because location is disqualifying, but because subcontracting you did not know about is a data boundary you have not evaluated and cannot supervise.
Ask insteadList every entity and subcontractor that will touch my data, the access each one holds, and the business associate agreements behind them.No exit terms, or exit terms that are worse than the entry terms. The moment you cannot leave cheaply, every subsequent conversation about performance happens from a weaker position than it should.
Ask insteadWhat is my notice period, who works the claims in flight on the day I give notice, in what format do I get my data back, and when is it destroyed?These are patterns worth probing, not accusations. A bidder who answers all six replacement questions directly has told you more about how the engagement will run than any proposal document will. The same six questions are fair to ask us, and the prior authorization command center whitepaper is one place we put our own operating model in writing so it can be checked rather than asserted.
Exhibit 5 · The 90-day migration
Open revenue is most at risk during the change itself.
The claims already in flight on cutover day belong to nobody by default, and that is where migrations lose money. Phases overlap on purpose: access work starts before the inventory is finished, and parallel running starts before access is fully closed out.
↔ Swipe the diagram
Ninety days is a planning frame, not a promise. The actual duration depends on the number of payers and clinicians involved, the state of your enrollment records, how long portal and clearinghouse access takes each payer to grant, and how much legacy AR is in scope. Phases that overlap in the diagram overlap in practice too, which is exactly why every open work type needs a named pre-cutover and post-cutover owner rather than a phase label.
Exhibit 6 · The questions
Fifteen questions. Ask every bidder the same fifteen.
Take the answers verbatim rather than in summary, and score them against Exhibit 1. The value is not in any single question, it is in asking identical questions of every bidder so the differences are theirs and not an artefact of how you asked.
-
Before you discuss price
Scope and ownership
- Walk me through every revenue-cycle process and tell me whether it is yours, mine, or shared.
- Who owns clinician enrollment and effective dates, and what happens to a session booked before a payer is live?
- Who obtains authorizations, tracks concurrent review dates, and escalates when a review date is approaching?
- Who works patient balances, and who talks to the client when they dispute one?
- Is legacy accounts receivable in scope, and if so from what date and at what fee?
-
Before you sign
Evidence and control
- Show me a de-identified denial work queue as it stands today, not a screenshot from a deck.
- Show me five rows from your live payer matrix with source, effective date, review date and owner populated.
- Give me one example of a denial reason you eliminated upstream, and tell me what you changed to do it.
- What can I see myself, on demand, without asking you for a report?
- List every entity and subcontractor that will touch my data, and the access each one holds.
-
Before you commit
Transition and exit
- Who works the claims already in flight on cutover day, and how is that recorded?
- How will we reconcile the accounts receivable position at cutover, and who signs it?
- What is the escalation path when I think something is wrong, and what response standard sits behind it?
- What is my notice period, and in what format do I get my data back?
- When my data is returned, when is your copy destroyed, and how is that confirmed to me?
These questions are as fair to ask us as they are to ask anyone else, and the answers should be in writing before a contract rather than discovered afterward. If you want the same fifteen answered against your own de-identified process map, send it to a senior partner, and if the honest conclusion is that you should keep billing in-house, Exhibit 2 above is where that conclusion is set out.
Mental health billing company FAQ
Questions to settle before selection.
The right answer depends on the practice, but the contract and transition documents should make each answer explicit.
What should a practice look for in a mental health billing company?
Look for a defined service scope, payer and clinician controls, named denial ownership, transparent work queues, source-dated rule management, secure data boundaries, measurable governance, and a transition plan that reconciles open claims and accounts receivable.
How can a practice compare mental health billing company proposals?
Normalize each proposal into the same process map. Identify who owns enrollment, benefits, authorization, claim edits, payment posting, denials, appeals, patient balances, reporting, and legacy accounts receivable. Then compare deliverables, exclusions, response standards, data access, and exit terms.
What should be included in a mental health billing transition?
A transition should inventory payers, clinicians, locations, enrollments, authorizations, interfaces, clearinghouse settings, open claims, denials, unapplied payments, patient balances, and accounts receivable. Each item needs an owner, cutover date, evidence source, and reconciliation method.
Should a billing company promise a universal financial outcome?
No single result can be assumed across practices. Performance depends on the starting backlog, payer mix, contracts, clinician enrollment, documentation, systems, service mix, patient-responsibility policy, and the scope assigned to the billing company. Targets should be based on a verified baseline and agreed measurement method.
Continue the research
Validate the buyer checklist against the service workflow.
This page owns the mental health billing company evaluation intent. Use the connected service and operations pages to inspect the actual workflows a proposal should cover.
Primary sources
Verify regulated workflows with official guidance.
These official sources illustrate two areas a mental health billing company may need to operationalize correctly. The exact payer and program matrix must still be validated for the practice.
Compare operating models with the same questions.
Share a de-identified process map or proposal scope. A senior partner can help identify missing owners, exclusions, control gaps, and transition risks before the decision is final.
Do not send protected health information through the public contact form.