Ask three billing vendors what outsourcing costs and you'll get three pricing structures that resist comparison on purpose. Here's the translation layer: the three models, the math for comparing them against your actual numbers, and the costs that don't appear on any proposal.
Model one: percentage of collections
The traditional billing-company structure: the vendor keeps a percentage of what they collect, commonly in the mid-single digits for eye care volumes. The pitch is alignment — they earn when you do — and the arithmetic is easy: practice collecting $1M annually pays roughly $40–70K a year at typical rates.
What to scrutinize: what counts as collected. Some agreements take their percentage of everything that flows in — including the vision-plan payments and front-desk copays your team effectively collected without vendor effort. That single definition can swing the real price by a third. Also note the structural quirk: the fee scales with your growth forever, while the vendor's marginal work doesn't.
Model two: flat monthly fee
A fixed retainer scoped to your claim volume, typically somewhere between the cost of a part-time and full-time hire depending on practice size. Predictable, budget-friendly, and immune to the collected-collections definitional games.
What to scrutinize: scope creep clauses and effort drift. Flat fees specify claim volumes — exceed them and the price steps up. And since the vendor's revenue doesn't rise with your collections, the alignment argument inverts: the incentive is to do the contracted minimum. Reporting discipline — weekly numbers, denial rates, AR aging — is how you keep a flat-fee vendor honest, so a vendor who resists reporting has answered your real question.
Model three: the dedicated remote billing specialist
The staffing model: one trained person, full- or part-time, working exclusively in your PM system at a flat monthly rate — typically in the range of $1,500–2,500 monthly for full-time through a healthcare staffing provider, training and backup included. Not a vendor relationship; a team member who happens to work remotely.
The economics differ from both service models in kind, not just degree: the rate is decoupled from your collections (no percentage drag as you grow), the work happens transparently in your own system (no black box, no lock-in), and the person carries your whole billing portfolio — scrubbing, posting, denials, AR, patient calls — rather than the slice a service contract defines. The tradeoff: it's a staffing relationship, with onboarding and management attention, and it's covered honestly in our billing-assistant guide.
The break-even math
Run your own comparison in four lines. Annual collections × percentage rate = the service-model price. Compare against the flat-fee quote at your volume, and against a dedicated specialist's annual cost. Then — the line everyone skips — estimate the recovery upside: if your current denial rate and aged AR reflect under-staffed billing, a competent operation typically recovers low-single-digit percentages of gross collections that were previously leaking. For many practices, that recovery alone exceeds the price gap between any two models, which reframes the decision: you're not choosing the cheapest vendor, you're choosing which model most reliably captures the leak.
The costs no proposal mentions
Whichever model: transition months run slower (parallel-running the old process is wise); your team still owns front-end accuracy — no billing arrangement survives bad demographics and skipped verification; and exit costs vary enormously — billing done in your system exits cleanly, billing done in a vendor's system exits like a divorce. Ask about all three before signing, and weight the answers as heavily as the price.
The honest summary: percentage models buy convenience at a price that grows with you; flat fees buy predictability that requires supervision; dedicated specialists buy capacity and transparency at the cost of light management. Priced against your own leak — not against each other — the right answer usually becomes obvious.




.png)
.png)
.png)
