Ask a multi-specialty group practice manager how much does a prior authorization cost per submission and you will usually get one of two answers. Either a shrug and a guess, or a fast number pulled from a vendor deck. Neither survives contact with your own timesheets. The honest answer is that the cost is not a fee on an invoice; it is labor, spread thin across three desks, mostly invisible until you add it up. And when you do add it up, the number is large enough to fund a hire you have been putting off.
The Medical Group Management Association put a hard edge on this. In its polling, roughly 60% of practices report that three or more employees touch a single prior authorization before it is resolved. Not three people over a month of appeals. Three people on one routine request. That statistic is the whole problem in one line, because prior authorization is not intellectually hard work. It is coordination work, and coordination is exactly what fragments when you split it across roles.
The Real Answer to How Much a Prior Authorization Costs Per Submission
Let us build the number from the parts you can actually see. Time studies across the industry consistently land a manual prior authorization at around 12 to 14 minutes of active staff time. Take the middle, 13 minutes. Load a coordinator at a $22 hourly wage and add the 30% you pay in taxes, benefits, and overhead, and you are at roughly $28.60 fully loaded per hour, or about $0.48 a minute. Thirteen minutes of that is a bit over six dollars.
But active time is the smallest part of the story, because prior authorization work is never done by one person in one sitting. The 13 minutes is split into three or four touches by three or four people, and each of those people carries a switching cost. Every time the file lands on a new desk, someone re-reads what the last person did, re-opens the payer portal, and re-establishes context. Add the phone hold time your staff eat while calling payers for status, which routinely runs 15 to 40 minutes a call, and the fully loaded cost per authorization lands closer to $11 in pure labor for a clean one and far higher for anything that gets kicked back.
Now scale it. A mid-size multi-specialty group running 30 authorizations a day, 250 working days a year, is producing 7,500 authorizations annually. At $11 each that is $82,500 in labor. That figure is not a rounding error. It is a prior authorization full-time staff salary plus benefits, spent not on a person who owns outcomes but on friction distributed across a team that would rather be doing anything else.
Mapping the Three Desks a Single Request Crosses
The reason it costs what it costs is the hand-off chain, so map it honestly. A typical prior authorization in a group practice crosses three functional owners, and the boundaries between them are where the time and the errors live.
flowchart LR
A[Order placed<br/>by provider] --> B[Front desk<br/>captures demographics<br/>and referral]
B --> C[Clinical staff<br/>pulls notes<br/>and codes]
C --> D[Billing submits<br/>to payer portal]
D --> E{Payer<br/>response}
E -->|Pended| F[Billing calls<br/>for status]
F --> D
E -->|Denied| G[Clinical rework<br/>and appeal]
G --> D
E -->|Approved| H[Schedule<br/>the service]Desk one is intake. The front desk or a referral coordinator captures the patient's plan, member ID, referring provider, and the ordered service. If the plan on file is stale or the member ID has a typo, everything downstream fails, but intake rarely finds out; the error surfaces two desks later as a denial.
Desk two is clinical. A medical assistant or nurse pulls the chart, finds the documentation that justifies medical necessity, and attaches the right diagnosis and procedure codes. This is the step that actually requires judgment, and it is the step most often stalled, because the clinical person is also rooming patients and returning calls. The file waits.
Desk three is billing. A biller submits through the payer portal or fax, then owns the worst part: status follow-up. Pended requests get called on. Denials get routed back to clinical for rework and re-submission. The loop between billing and clinical can cycle two or three times on a single authorization, and each cycle is another day the patient is not scheduled.
Three desks, and no one of them owns the outcome. That is the structural flaw. The patient's authorization is everyone's responsibility, which is a polite way of saying it is no one's.
Where the Hand-Off Chain Actually Breaks
If you want prior authorization backlog help that lasts, you have to be precise about which joints in the chain fail, because they are not equally broken.
The first fracture is intake data quality. When eligibility and plan requirements are not verified at the moment the order is placed, bad data flows downstream and only reveals itself as a denial or a pend. A denial for "not a covered benefit" or "no authorization on file" is very often an intake miss dressed up as a payer decision. Fixing it means re-doing all three desks' work.
The second fracture is the clinical-to-billing gap. Clinical knows the medical story; billing knows the payer's rules. Neither fully speaks the other's language, so requests go out under-documented and come back pended for exactly the record clinical already had. The information existed; it just did not travel with the request.
The third fracture is status follow-up, and it is the single biggest time sink. A pended authorization is not resolved until someone calls the payer, waits on hold, navigates the IVR, and extracts a status. That call produces nothing a patient values. It is pure carrying cost, and it recurs for every request that does not clear on the first pass, which in many specialties is most of them.
Notice that none of these three fractures require clinical judgment. Verifying a benefit, carrying documentation from one system to another, and calling a payer for status are mechanical. That is precisely why they are automatable, and why automating them collapses the chain rather than just speeding it up.
Collapsing Three Roles Into One Supervised Queue
The move that changes the cost curve is not hiring a fourth person to manage the three. It is removing the mechanical work so one person can supervise the whole chain instead of three people each guarding a slice of it.
CallSphere Health attacks the three fractures directly. At intake, the AI front desk captures the patient's plan, member ID, and ordered service as structured data at the moment of the call or the order, and runs the benefit and requirement check right then, so a stale plan or a missing referral surfaces before it becomes a downstream denial. The clinical-to-billing gap narrows because the documentation and codes attached at intake travel with the request instead of being reassembled at desk three. And the biggest sink, status follow-up, moves off your staff entirely: the system's multilingual voice AI places the payer status calls, sits through the hold and the IVR, and returns a status your coordinator reads in seconds instead of dialing for.
Here is what that does to the org chart. Instead of one authorization touching three employees in sequence, it flows through an automated queue, and a single clinical coordinator supervises exceptions.
flowchart TD
A[Order placed] --> B[AI captures intake<br/>and checks benefits]
B --> C{Clean and<br/>authorized}
C -->|Yes| D[Auto submit<br/>with documentation]
C -->|Needs review| E[Coordinator<br/>handles exception]
E --> D
D --> F[AI follows up<br/>on status]
F --> G[Approved<br/>and scheduled]The three-desk chain becomes one queue with a human on the exceptions. The requests that used to consume intake, clinical, and billing in series now run themselves when they are clean, and your coordinator spends her day on the 20% that genuinely need a person, not the 80% that just needed a phone call made. You can see the specific capabilities behind this on the /features page, and the labor math is exactly why a subscription priced against a fraction of one salary pays for itself, which the /pricing breakdown lays out.
What Reclaiming a Coordinator's Day Is Worth
Return to the $82,500 a year our example group was spending. The point of collapsing the chain is not to eliminate the coordinator; it is to change what the coordinator does. When the mechanical steps run automatically, that same person stops being a relay between two other desks and becomes the owner of authorization outcomes, working exceptions, appeals, and the hard cases that actually move revenue.
The dollar logic works two ways. On the cost side, you stop paying three fully loaded people to each handle a third of every request; the switching cost between desks disappears, and the hold-time hours evaporate. On the revenue side, faster and cleaner authorizations mean fewer services delayed past the point where the patient reschedules or walks, and fewer denials that turn into write-offs because someone submitted late or short on documentation. For a multi-specialty group, a service that gets authorized and scheduled this week instead of pended for two is real money that would otherwise leak out the bottom of the funnel.
There is also a retention dividend that never shows up in the authorization line item. The staff who spend their days on hold with payers are the same staff who burn out and quit, and every departure costs you months of hiring and training. Take the worst work off their plate and you keep the people who know your payers and your patients. That is prior authorization backlog help that compounds, because the coordinator who stays gets better at the exceptions every quarter.
The Number Worth Putting on the Board
So, how much does a prior authorization cost per submission? For a clean one at your practice, somewhere around $11 in labor. For a pended or denied one, several times that. Across a year, roughly the salary and benefits of a full-time coordinator you are already paying, just spread invisibly across three desks so no line item ever shows it.
The MGMA finding that 60% of practices put three or more people on a single authorization is not a story about lazy staff or bad process discipline. It is a story about a chain that fragments coordination work across roles and then pays the switching cost on every request. Map your own chain, count the desks, and price the hold time honestly. Then decide whether that money is better spent on a phone queue or on the person who could be closing your hardest cases. The number, once you see it, tends to make the decision for you.