Prepared for Dan Neissany · for the practice you described on SundayThe practice runs on people remembering.
Here is the layer that remembers for them.
A 30-person outpatient orthopedic PT practice, CareCloud at the center, 30 new patients a week, 500 claims a week. You coach Leadership, Culture and Strategy. This is the Operations layer: your macro flow, your acceptance criteria and your job aids, running on top of CareCloud, in the order we would build it, with the numbers from our call.
The preview is the same practice, running. Every view is clickable. All names and rows are fictional.
The math, then what we heard
Your numbers turned into hours and dollars, live, then the practice as you played it and the sentence that names the real problem.
Four Ops, in order
Front Door, Follow-Through, the Denial Workbench, then the habit rail. Each with an honest alternative.
Investment and the after
What it costs, what twelve months looks like, and how we start.
This is your five-step method (map the macro flow, assign owners, set acceptance criteria, attach job aids, train and certify) with a system underneath it that keeps the criteria true between coaching sessions. The preview shows the same six stages live on its Today view.
The Math: your numbers, in dollars
You asked for a real number, so it comes first. The calculator has your inputs already in it; move anything you disagree with and the results follow. The loaded hourly cost is our assumption: a front-desk salary of about $38,700 a year (Salary.com, June 2026) lands near $19 an hour before benefits and overhead, and billing staff run higher, so $30 is a round, conservative loaded figure. Everything behind these numbers, and what we would build about them, is in the sections that follow.
Assumptions, all editable above: 50 working weeks; the 8% PT average is the StrataPT benchmark across PT, OT and speech practices (vendor data, refreshed daily); MGMA puts all-specialty first-submission denials at 7 to 8%; calls that become a started plan ≈ new patients ÷ 0.7 answered ÷ 0.8 complete; the "1 in 10 never calls back" share is deliberately conservative (CallRail's 2025 survey puts the share who call a competitor after an unanswered call far higher, but that is a consumer survey, not a clinic study). None of this counts the denials Front Door prevents, or the plans of care Follow-Through finishes. Those are real and we would rather show them on a live practice than estimate them here.
Where You Are: a practice that is good at the clinical part
You described an independent outpatient orthopedic PT practice, about 30 people, where the owner still treats. A patient calls, the front desk takes their information, someone at the desk or in the small in-house call center verifies insurance by phone or portal, the patient gets a callback with benefits, comes in for an evaluation, and the clinician schedules the plan of care. Notes go into CareCloud, billing submits from CareCloud, and the patient gets one follow-up call after the first visit. Salesforce handles marketing. A separate dashboard tracks cancellations. You said that stack covers 99% of what you need, and nothing is on the wishlist.
The numbers you gave: verification runs 2 to 40 minutes per patient, 30 patients a week. About 500 claims a week, 15% denied, 45 to 60 minutes to work each one. The worst part of a denial is not the work, it is reaching a human at the insurer, sometimes 45 minutes on hold, even with batch calling and portals already in use.
The softer things: clinician notes run late, mostly because people don't use their own schedule well, and that spills into evenings and stresses billing. The habits you have trained (walking patients out, watching exercise, checking the waiting room) need re-teaching. Written scheduling strategies exist and are not followed. You are proudest of patient results, and they show up as referrals and reviews.
On Tuesday you added the two numbers that matter most for a front desk: practices miss about 30% of their calls, and of the ones they answer, about 20% don't capture everything needed, which turns into denials and clinicians who walk into an eval without the picture.
"If I knew, honestly, I would work on a solution."
On what stops communication and adoption from sticking · discovery call, Sunday
The Bottleneck: nothing in the practice enforces what the practice already knows
The strategies are written. The training happened. The scripts exist. What is missing is the thing that makes them happen on a Tuesday at 4 PM when the phone rings and the front desk is short a person. Every pain you named is the same pain: a process that lives in people's memory, and memory under load.
Three things are true at the same time:
- The phone leaks, and the leak shows up as denials. Roughly a third of calls go unanswered and a fifth of the answered ones are incomplete. Across medical practices, about 38% of denials start at registration, eligibility or authorization (Change Healthcare data via MGMA). An incomplete intake is a denial six weeks later.
- The denial rate is about twice the norm for PT. 15% against an 8% average across PT, OT and speech practices (StrataPT benchmark) and 7 to 8% across all medical groups (MGMA, 2026). 75 denials a week at 45 to 60 minutes each is 56 to 75 staff-hours, the work of one and a half to two full-time people, most of it waiting.
- What is taught does not stick. Late notes, unwalked patients, the empty waiting-room check. You can coach it, and you do. Coaching without a system that reminds, records and escalates is coaching on repeat.
Missed calls and incomplete intake at the front door. Referrals and reviews are strong but untracked.
One follow-up call per plan of care. Drop-off is caught late. Notes and habits depend on the person.
15% denials, an hour each, and a hold queue you can't control. The biggest hour count in the building.
What We Looked At: the layer between the phone and CareCloud
CareCloud stays. Salesforce stays. The metrics dashboard stays. A Practice OS sits in the gap those three leave: the phone and texting are not connected to the schedule, follow-up after a visit is manual, denial work happens in inboxes and on hold, and owner-level reporting gets rebuilt by hand. That is the same gap in every specialty we reviewed, from WebPT and Prompt in PT to Dentrix in dental to athenahealth in medical.
The categories we compared before recommending anything: patient-communication platforms (Weave, Solutionreach), medical answering services, AI phone agents built for clinics, denial-management and clearinghouse tools (Waystar, Availity Essentials), outsourced revenue-cycle firms, and CareCloud's own add-ons (Breeze for Therapy, its stratusAI phone agent and cirrusAI appeal drafting, which a practice may already be paying for). Each Op below carries an Ideal path and an honest Alternative from those categories, so a practice can take the recommendation or the cheaper road and still get most of the value. What none of them do is tie the six stages together for the owner on the system the practice already runs.
Vetted tool selection, a sequenced rollout and the math behind it is normally the output of a mid-market operations consulting engagement. We productize it for practices that would never get quoted by those firms, and we build and run the result instead of handing over a deck.
The Diagnosis: three Drivers, four Ops
Each Op maps to one Driver from Section 03 and to the macro-flow stages it enforces. They are listed in build order, the same order as the Priority Stack in Section 06. Each one names its acceptance criteria and owner, says what happens today and what changes, and gives the two ways to get there.
Front Door
Stages: Lead, Intake · Acceptance criteria: 100% of calls answered with the eval offered on the same call; intake complete before anything books; benefits verified within 24 hours · Owner: front desk lead
Today about 30% of calls go unanswered and 20% of answered ones miss something, and that second number is where a share of the denials begins. Front Door answers every call, first ring, including before open and over lunch. It runs the full intake script every time (name, callback number, insurance and member ID, referring physician, authorization flag, reason for the visit in the patient's words), and the record cannot close with a blank. It starts the benefits check electronically at booking and again 48 hours before the first visit: an electronic eligibility check costs a practice about $2.00 against $8.57 by phone (CAQH Index, 2024), and it takes seconds instead of a morning. Then it books the evaluation into CareCloud's schedule. The front desk sees a complete record and a transcript, not a sticky note. Calls that need a human get a human, with the intake already done.
Because you told me to force this part: the same door tracks where every new patient came from (which physician, which search, which former patient) and asks for a review on discharge. That feed goes to Salesforce, so the marketing and sales work you are building has real source data instead of "how did you hear about us" guesses.
AI phone agent on the practice's own number, portal-first verification
- Overflow and after-hours first, then all calls once the staff trust it (usually week 2)
- The intake job aid runs on every call; nothing books until it is complete
- Electronic eligibility at booking and 48 h before visit 1; phone-only payers queued for staff with the hold batched
- Reminders at 3 days and 1 day (two reminders beat one in a 54,000-patient trial); waitlist offers fill cancelled slots
- Source tracking and review requests built in
CareCloud's own phone agent, or a human answering service
- CareCloud now sells an AI desk agent (stratusAI) that answers and schedules; ask what it captures, what it costs, and whether it runs an intake job aid
- An answering service covers after-hours and overflow; intake quality depends on its script and turnover
- Either way: no owner view, no benefits check tied to the plan, no source data to Salesforce
Follow-Through
Stages: Scheduling, Follow-up · Acceptance criteria: a touch after every visit; next visit booked before checkout; cancels and no-shows ≤ 12%; plan-of-care retention ≥ 80% · Owner: front desk lead, clinician on flags
This one is your idea from Tuesday, and it is the right one, with one adjustment the evidence forces. Today a patient gets one call after visit one. Follow-Through sends a short check-in after every visit: how did today go, anything to flag, here is your next appointment. Routine answers are logged. Anything non-routine (pain that got worse, a question about the exercises, a scheduling problem) goes to the treating clinician as a flag, not a voicemail. The adjustment: the every-visit touch is a text, and a human call happens on triggers. Cochrane's review found text reminders perform about the same as phone calls at roughly 60% of the cost, and nobody has published a controlled result for a human call after every visit. So the calls go where they count: after the first visit, after the first cancellation, after a gap of more than ten days, and after any reply that says something is wrong.
Two peer-reviewed numbers behind this. A PT patient who misses more than a fifth of visits in the first month is 3.5 times more likely to drop out entirely (JOSPT Open, 2024). And in a study of 445,000 PT patients, the strongest predictor of the next no-show was the number of previous cancellations (PLOS One, 2021), which is a scheduler field, not a clinical one. The first cancellation is the trigger. It also feeds your adoption problem from the other side: the patient-facing steps happen whether or not the front desk remembered.
Text after every visit, human call on triggers, recall on silence
- Text within two hours of checkout, two questions, generic wording, opt-out every time
- Human call on: visit 1, the first cancellation, a gap over 10 days, a "no" reply
- Non-routine replies flagged to the clinician's queue the same day
- Missed or cancelled visit → rebooking offer and a same-day callback task; two silences → recall
- Review request on discharge, tied to the referral tracking in Op 1
Front desk calls after visit one and visit four
- What you do today, extended to a second checkpoint
- Costs staff time at the busiest hours of the day
- Nothing catches the patient who quietly stops after visit six
Denial Workbench
Stage: Billing · Acceptance criteria: clean claims ≥ 97% (your number); denial rate ≤ 8%; every denial triaged within 48 hours; nothing ages out unworked · Owner: billing lead
Start with the gap: 15% denied, against an 8% average for PT practices and 7 to 8% across medical groups. The first thing we build is not software; it is a 90-day pull of your remittances rolled up by reason code and payer, so the practice knows whether its 15% is eligibility, authorization, documentation or coding. Everything after that is aimed at what the rollup shows.
Honest second: the 45 minutes on hold with the insurer is outside our control, and we will not pretend otherwise. What the Workbench removes is everything around the hold. Today a denial lands, someone reads it, decides whether it is coding or documentation, chases the provider for a note, drafts the appeal, and calls the payer. The Workbench reads every denial as it arrives from CareCloud, sorts it by reason code, sends the documentation request to the right clinician with the due date attached, drafts the appeal from the note and the payer's own rules, and batches the calls that still need a human by payer so one hold covers ten claims. Every step is logged with a timestamp, so the owner sees where the time goes instead of guessing.
The second effect is upstream. The 20% of intakes that miss something are a share of the 15% that get denied. Front Door closes that gap at the source, so the denial rate itself comes down over the first quarter.
OS workbench on top of CareCloud's claim responses and the payer portals
- Triage by reason code within minutes of the remittance; sorted by dollars and days to deadline
- Documentation requests routed to the clinician with a due date; nudges until the note is in
- Prevention counters that follow the rules: progress report by visit 10, recert by day 90, the Medicare therapy threshold ($2,480 in 2026) with the KX modifier prompted, CQ/CO on assistant visits, authorizations flagged at three visits left
- Claim status pulled electronically (7 minutes versus 25 by phone, CAQH 2024) so calls are for disputes, not questions
- Appeal drafts for review; the billing lead sends, never the system
- Hold time and resolution time measured per payer, so the practice can see which contracts cost the most to keep
CareCloud's appeal drafting, or outsourced billing
- CareCloud's cirrusAI drafts appeal letters; ask whether it is in the contract and what it does before the letter (triage, routing, aging)
- A billing firm works the claims for 3.5 to 8% of collections (published and reported PT rates, 2026); it removes the hours, not the causes, and intake gaps keep generating denials
- Reporting is theirs, on their schedule
Habit rail
Stage: Visit · Acceptance criteria: notes signed same day ≥ 90%, per clinician; next visit booked before the patient leaves; the 10-chart spot audit run every week without anyone remembering to · Owner: lead clinician, with the office manager as champion
This is the answer to the magic-wand question, and it comes last on purpose, because it only works once the first three Ops have taught the team that the system is on their side. The research on why trained behavior fades is sobering and useful. Written procedures with no measurement produce reported compliance of 92 to 98% and no change in what actually happens (a 101-hospital checklist study). People do the behavior 78% of the time when they know they are watched and 55% when they don't. Feedback moves behavior when it is about the individual, compared to the best peers, delivered by someone who has a relationship with them, and paired with one specific action; repeating the same feedback makes it weaker. So the rail is built from those findings, not from reminders.
What that looks like: the habits you keep re-teaching become receipts the OS records passively, not boxes anyone ticks. A note not signed by end of day is a timestamp, not a confession. A patient checked in and not brought back in twelve minutes is a clock on the front-desk screen. A discharge without the next step booked is a missing field. Each person gets one weekly page with their own numbers next to the top quartile and one suggested action. The first miss goes to the lead, the second to you. The 10-chart spot audit you already recommend runs itself every week and posts its score. None of it replaces your coaching. It is the part of coaching that happens when you are not in the room, and for the first time it is measured. You handle Leadership, Culture and Strategy. This is the O.
Flagged for development after the first three Ops are live. Scope depends on what CareCloud exposes for note status and check-in events; we confirm that during build.
The Priority Stack: start here, in this order
- 1Front DoorWeeks 1–2Fastest proof, no clinical workflow touched, and it starts shrinking the denial rate from day one.
- 2Follow-ThroughWeeks 2–3Rides the consent and texting rails Front Door sets up. Your idea, visible to every patient by week three.
- 3Denial WorkbenchWeeks 3–6The biggest hour count, and the biggest gap to the benchmark. Starts with the 90-day remittance rollup; needs claim status connected, which is the longest single setup step.
- 4Habit railAfterOnce the team trusts the first three. Scoped with the owner, one habit at a time.
How It Runs: four rules that make this different from another tool
- Your method is the spec. Macro flow, micro steps with owners, acceptance criteria, job aids, certification. The OS holds the criteria as guardrails, attaches the job aid at the moment of work (the intake script on the call, the checkout checklist at checkout), and runs the spot audit on a schedule. Your weekly stand-up opens on the six-stage strip instead of on memory.
- Nothing clinical leaves CareCloud. The OS works from appointment status, claim status and call events. Notes, diagnoses and plans of care stay where they are. That keeps the automation layer out of most of the PHI exposure and keeps the BAA simple.
- Every action has a receipt. A ✦ mark and a timestamp on anything the system did, in a feed the owner can read in two minutes each morning. Adoption stops being a feeling; the single point of failure stops being a person.
- The practice sets the leash. Green actions run on their own (answering, reminders, check-ins). Yellow actions are drafted and wait for a person (appeals, patient balance notes). Red actions never run without a click (money, anything that touches the record). The preview shows the switches.
And the part that matters to you specifically: we build and run it. Your LOCS work sets the behavior; the OS is the machinery that makes it stick between sessions. Neither one replaces the other.
Investment: what the engagement costs
One number, sized to the practice, no per-seat line items. For practices up to about 20 active users the published shape is a setup of $7,500 to $10,000 and $800 to $1,500 a month, twelve-month initial term, then month to month. A 30-person practice is sized individually inside that same shape, one flat monthly number, and it needs single sign-on against the practice's Google or Microsoft directory so an ex-employee's access ends the day their account does. The setup covers the build of all three Ops; the monthly covers running it: the answering, the check-ins, the triage, the monitoring, the changes as the practice changes.
For context. The denial desk alone is one and a half to two full-time people. A front-desk hire is about $38,700 a year before benefits, and the front desk is the highest-turnover seat in the building. The monthly lands at roughly one answering service plus one patient-communication tool, and well under one salary.
Never a hostage. The practice's data is exportable in standard formats on request, at no fee, cancellation included. The price is fixed for the term. If they ever want to own the system outright, that price is printed in the agreement, not negotiated later. If they leave, they get thirty days of transition help.
Anything that stores or moves protected health information for the practice needs a Business Associate Agreement; we sign one. The design above keeps clinical data in CareCloud and triggers on appointment and claim status, not on diagnoses. Patient texts need consent at intake, generic wording, a sender name and an opt-out every time. The TCPA rules around automated texting became less settled after the Supreme Court's 2025 McLaughlin decision, not more. Before any of this is sold to a practice, a healthcare attorney reviews the consent language and the BAA. This roadmap is not legal advice.
The After: what twelve months looks like
The phone. Nobody at the practice remembers the last time a call went to voicemail. New patients arrive with a complete record and a benefits check already run. The front desk starts its day from a list, not from a backlog.
The desk. Denials arrive sorted. The billing lead reviews appeal drafts instead of writing them, and makes payer calls in batches with a purpose. The owner's morning view shows which payer cost the most hours last month, which turns into a contract conversation instead of a complaint.
The clinician's evening. A note that is not signed by five gets a nudge, not a lecture. Patients who are struggling say so in a text at 6 PM and the clinician knows before the next visit. Fewer surprises, fewer notes done at the kitchen table.
The owner's month. The conversations that used to repeat once a month have a receipt behind them: here is what the system caught, here is what the team did with it. Coaching turns into calibration.
Financial impact. Using the Section 01 inputs as they stand, the denial desk alone is worth about — a year in staff time; halving the handling around the hold recovers about —, and the missed-call number puts roughly — a year of plans of care back in play. Against a monthly in the low four figures, the return is there in the first quarter, before counting a single prevented denial.
And the L, C and S. The practice you coach on leadership, culture and strategy has an operations layer that holds the line between sessions. That is a different offer than coaching alone, and it is one you can put in front of every LOCS client without building a software arm.
What Happens Next
- Poke the preview. The Practice OS preview is this practice, running. Click into Messages for the 7:12 AM call, Claims & money for the denial batch, and Today for the switches that set the leash.
- Thirty minutes when you are back. Walk the four Ops, argue with the calculator, tell me which parts your clients would recognize and which they would not.
- One question for CareCloud before any build. Whether appointment status and claim status can be exported or fed out of the practice's CareCloud product, and at what cost. Its public interface covers clinical records only. If the answer is no, claim status comes through the practice's own clearinghouse and scheduling through a nightly export. We ask in writing; the answer shapes weeks 1 to 2.
- Pick the first practice. This preview is the free one. The first real practice is a paid pilot at the terms above, built on their own numbers, and it becomes the reference every LOCS client after it gets to see.
- How the two of us work. The practice contracts with ADVANCE.DESIGN; you bring the practice and the L, C and S. We settle the shape of that on the call, in writing after.
This roadmap is yours to keep, and it will be updated here as the thinking changes.