Inventory Management
Procurement
Method Pay
Analytics & Reporting
The most expensive employee in a DSO finance department is rarely the controller. More often, it is the AP clerk who keys 200 invoices a day from memory.
Their salary has nothing to do with it. The cost lives in what gets through. The duplicate that looks like a backorder. The unit price that crept up two cents per box. The invoice for ten units when seven arrived. None of it gets caught at speed, because catching it at speed sits beyond human capability. Systems handle that work.
Finance leaders typically evaluate AP automation as a cost-reduction play. Fewer hours keyed. Fewer errors are caught downstream. A leaner team. The framing holds up, as far as it goes, but it sells short what becomes possible at elite DSO scale. Something more structural happens when invoices stop being a human task. AP shifts from a processing function to a control function. The operating model changes, along with the cost line.
This piece is about what that change looks like in practice, and why getting there takes more than turning automation on.
It’s worth establishing what the current state actually costs before getting into what changes.
Ardent Partners benchmarks the per-invoice processing cost of best-in-class AP teams at $2.78. Average organizations spend $12.88. Fully manual teams spend $19.83 or more. That 78% gap between top quartile and bottom quartile compounds with every invoice an organization processes.
At elite DSO volume, the difference between a best-in-class operation and a manual one runs into hundreds of thousands of dollars annually, in pure processing cost alone.
Errors do not just cost time to correct. They cost money. APQC benchmarking shows that 0.8% to 2% of total disbursements at high-volume payers are lost to duplicate or erroneous payments, and IOFM estimates the cost to investigate and correct a single erroneous payment at $53.
For a DSO processing tens of thousands of supply invoices a year, the math is unambiguous. Manual AP is labor plus errors plus leakage, and at elite scale, all three feed each other. The leakage line is the one that quietly drains margin without showing up on anyone's dashboard.
It is common for DSOs to "solve" AP by routing supplier payments to a credit card on autopay. It looks efficient. The invoice arrives, the card gets charged, no one keys anything, the staff hours disappear.
It is also the most expensive way to operate.
Autopay does not eliminate AP work. It eliminates AP verification. There is no three-way match. There is no line-item check. There is no catch on duplicate invoices, no flag when a supplier quietly raises a unit price, no protection against billing for goods that were never received. The grand total goes through, and every piece of detail AP would have used to protect the organization disappears with it.
That is the trap to recognize before evaluating any AP solution. Automation that strips out the labor by stripping out the controls is a worse system than the manual process it replaced. The benchmark to use measures verification preserved, not human time removed.
Three-way match is the foundational control of any serious AP function. It compares three documents before a payment is approved: the purchase order, the receiving record, and the supplier invoice. If all three agree, the bill moves forward. If they do not, the discrepancy gets investigated.
Dental procurement platforms that offer any form of invoice matching often perform it at the order level. The PO total matches the invoice total, so the system says everything is fine. That is a blunt check, and it misses the variances that actually cost DSOs money. A supplier can invoice at a higher unit price on one line and a lower price on another, and the totals reconcile. A location can receive nine of ten units, and order-level matching never catches it.
MethodPay performs three-way matching at the individual line item level. Every product on every invoice gets checked for quantity, unit price, and extended price against the PO and the receiving record.
If a supplier invoices at a higher unit price than was on the PO, it gets flagged. If four units were ordered and invoiced but only three were received, it gets flagged. Backorders, split shipments, and multiple invoices against the same PO are aggregated into a unified view of where each PO actually stands. Duplicate invoices are caught automatically.
The reason this matters goes beyond pricing accuracy. The 2026 AFP Payments Fraud and Control Survey found that 76% of organizations reported attempted or actual payment fraud in 2025. Line-item three-way matching is one of the few systematic protections against paying for things the organization did not agree to buy, or did not receive. It functions as a financial control, and at elite DSO scale, it prevents the slow leakage nobody notices until year-end.
When set up correctly, the sequence looks like this:
An invoice arrives via email, sent directly to Method by the supplier. The platform reads the invoice and extracts every line item, quantity, unit price, extended amount, coding each one for processing. Three-way match runs automatically against the PO and the receiving record. You’ll either get indication that the match is clean or there is a discrepancy that needs attention.
If the match is good to go and the invoice total falls below a defined dollar threshold, the bill auto-approves and routes directly to the accounting system. It arrives fully populated with the expense account, department, and location, with the original invoice document attached. If the invoice exceeds the threshold, or it comes from a supplier flagged for additional review, the bill routes to an approver with the discrepancy surfaced for resolution.
What this describes is exception-based AP. Humans only touch invoices that require human judgment. From there, it’s important that there’s a control layer to ensure the right people are involved.
Cost savings get a CFO's attention. The control layer survives an audit, a recapitalization event, or a due diligence cycle. Three controls become enforceable at scale once invoices flow through a single system.
No single person should ever be able to buy, receive, and pay for goods. That is a foundational internal control, and at elite DSO scale, the financial exposure of getting it wrong is significant. Method's role and permission structure (12+ distinct permissions, with location-level role assignment) makes segregation of duties consistently enforceable across every location, without adding headcount. The buyer cannot get into receiving. Receiving cannot get into buying. Neither can get into AP.
Every action on every bill is logged: who touched it, what changed, when, and why. Internal notes between processors and approvers stay part of the record. The original invoice document is attached automatically to the accounting system entry. Audit prep stops being a multi-week reconstruction effort.
The same workflow engine that handles purchasing approvals handles AP approvals. Bills above a threshold, bills with discrepancies, bills from specific suppliers, all route to the right reviewer based on configurable logic. The CFO does not have to look at every invoice. The CFO has to look at the ones that matter.
Four things change at the level of how the DSO actually operates when invoices process themselves.
Headcount decisions change. AP scales sublinearly with location count. Adding the 51st practice does not add an AP FTE. Adding the 75th practice does not double the team.
They stop keying data and start managing exceptions, supplier disputes, and recoveries. That work is higher-leverage, harder to outsource, and where finance actually adds value.
PO to payment, all in one system with no double entry between procurement and AP or reconciliation between disconnected tools. Spend visibility and AP visibility become the same thing.
Payment terms (Net 30, 60, 90) get managed as a strategic lever rather than bypassed by autopay charges that clear the moment an invoice lands.
The question elite DSOs should be asking is whether the AP function the organization has today can still be the AP function it has at 75 locations, or 100.
If the answer is no, automation becomes a prerequisite rather than an optimization.
