|
The short answer: AP automation pricing is set by the model, not the headline rate. Platforms use five: per-invoice (a fee for each invoice processed, usually with volume bands), flat subscription (a fixed fee for a defined invoice range), per-user or per-seat licensing, transaction-based pricing on payments, and hybrid models combining a platform fee with per-invoice or per-payment charges. Each behaves differently as you scale — per-user pricing punishes wide approval chains, transaction pricing scales with payment volume, and subscription pricing wastes money below the band and triggers overage above it. Total AP automation cost also includes three layers rarely shown on a pricing page: transaction fees, implementation, and features gated behind higher tiers.
|
Most pricing comparisons stop at the sticker. That is the least useful number in the contract, because two vendors quoting the same annual figure can diverge by a wide margin once your approver count doubles or you add an entity. What actually determines AP automation cost is the shape of the model — which variable your bill is indexed to, and whether that variable is one you control.
The five AP automation pricing models
|
Model
|
What you pay for
|
Scales with
|
Watch for
|
|
Per-invoice
|
A fixed fee per invoice processed, usually tiered so unit cost falls as volume rises
|
Invoice volume
|
What counts as an invoice — credit notes, multi-page documents, rejected items
|
|
Flat subscription
|
A fixed monthly or annual fee covering a defined invoice band or feature bundle
|
Nothing, inside the band
|
Overage rules, and paying for headroom you never use
|
|
Per-user / per-seat
|
Each person with access — clerks, approvers, administrators
|
Headcount touching AP
|
Whether approvers, auditors and read-only roles are billed seats
|
|
Transaction-based
|
Fees on payment execution: ACH, cheque, wire, virtual card, FX
|
Payment volume and method
|
FX markup and interchange, which rarely appear on the pricing page
|
|
Hybrid
|
A platform fee plus per-invoice or per-payment charges
|
Two or more variables at once
|
Compounding — a volume spike raises both layers simultaneously
|
A sixth model, outcome- or savings-based pricing, ties part of the fee to measurable results such as touchless processing rate or captured early-payment discounts. It is discussed more than it is sold, and where it exists it usually applies to a slice of an enterprise contract rather than the whole thing.
Why per-user pricing behaves differently from the rest
Four of the five models index to something that correlates with the value you receive. Per-user pricing indexes to how many people your approval policy involves — which is a governance decision, not a usage one.
The effect is that good financial control becomes expensive. A business that routes invoices to the budget owner who actually incurred the spend needs more approvers than one that funnels everything through a single finance manager. Under seat-based pricing, the better-controlled organisation pays more. Seat-based vendors commonly publish tiers in the $45 to $89 per user per month band, though published figures vary noticeably between resellers and comparison sites, and approver-only rates are often restricted to higher tiers.
The practical test when evaluating a per-user quote: count every person who would ever need to open an invoice — approvers, delegates, auditors, procurement, occasional cover during holidays — and price that number, not the size of the AP team. Multi-site operators feel this hardest. Applegreen runs AP across 193 retail sites and Gold Care Homes across 29 UK sites, both documented in our customer stories — under seat-based pricing, that site count becomes a recurring bill.
The four layers of AP automation cost most buyers miss
The published model is one layer of four. Comparing vendors on the first layer alone is the single most common evaluation error.
- Platform or processing fee — the number on the pricing page. Per-invoice, per-seat, or a flat band.
- Transaction fees — per-payment charges for ACH, cheque, wire, card and international transfer. On payment-led platforms these can exceed the software fee. Card acceptance and FX markup are usually percentage-based, so they grow with spend rather than invoice count.
- Implementation and integration — one-time setup covering process design, ERP connectors, data migration, testing and training. Rarely published, and materially larger for multi-entity or multi-ERP deployments.
- Modules and tiers — features gated behind an upgrade: multi-entity, multi-currency, PO matching, supplier portal, fraud detection, analytics, e-invoicing compliance. The functionality you assumed was included is often the reason the quote moves.
- Your current cost per invoice, calculated as above and including the costs finance teams routinely omit — query handling, late payment charges, reprocessing.
- A defensible target cost per invoice, not the vendor’s best-case figure. Use the gap between your baseline and the best-in-class benchmark as the range, and model the conservative end.
- Total cost of ownership across the full term, covering all four cost layers over three years rather than year-one licence fees. Implementation is a one-off; per-payment fees compound.
- The non-financial case, which is usually what actually wins approval: faster month-end close, audit readiness, duplicate payment prevention, and capacity released without adding headcount.
|
Total cost of ownership
Annual TCO = platform or per-invoice fees + transaction fees + amortised implementation + module upgrades + internal effort still required (exception handling, rekeying, supplier queries).
That last term is the one buyers leave out, and it is the one automation is supposed to remove. A cheaper platform that leaves 30% of invoices to be worked manually is not cheaper.
|
Which model fits which AP profile
|
If your AP looks like this
|
Model that usually fits
|
Reasoning
|
|
Low or unpredictable volume, growing fast
|
Per-invoice with volume bands
|
Cost tracks activity; no wasted capacity while you scale into it
|
|
Steady, high volume, stable headcount
|
Flat subscription or negotiated band
|
Predictable budget; unit economics improve inside the band
|
|
Small centralised team, few approvers
|
Per-user is survivable
|
Seat count stays low enough that the model does not distort
|
|
Wide approval chains across departments or sites
|
Avoid per-user
|
Every additional budget owner adds recurring cost for occasional use
|
|
Heavy international or card payment volume
|
Scrutinise transaction pricing
|
FX markup and interchange can dominate total cost
|
|
Multi-entity or multi-ERP
|
Expect a higher implementation layer
|
Configuration and mapping effort, not licensing, is the driver
|
Sector shapes the answer more than company size does. Retail and construction operators carry high site counts and wide approval chains, so per-user models get expensive fast. Manufacturing tends to be PO-heavy and high-volume, which favours per-invoice bands. Private equity portfolios and multi-academy trusts in education add entities faster than invoice volume, so per-entity charges matter more than the unit rate. Charities and healthcare providers typically need audit depth included rather than gated behind a tier.
Calculate your cost per invoice before you compare quotes
No vendor quote is meaningful without a baseline, and most finance teams do not have one. The calculation is simple:
|
The formula
Cost per invoice = total AP costs over a period ÷ total invoices processed in that period
Include: labour and overhead, existing software, scanning and postage, supplier query handling, late payment charges, and reprocessing caused by errors. Excluding the last three is why internal estimates usually come in low.
|
For an external reference point, Ardent Partners’ Accounts Payable Metrics That Matter in 2025 puts the average cost to process a single invoice at $9.40 against $2.78 for best-in-class performers, with average processing time of 9.2 days versus 3.1 days. Cross-industry benchmarks from APQC are useful for sense-checking your own figure once you have it. Our invoice processing savings calculator will give you a starting estimate against your own volumes.
Turning the numbers into an AP automation business case
A pricing comparison is not a business case. The business case is what survives contact with a CFO, and it needs four things the vendor quote will not give you:
AP automation ROI typically arrives through lower cost per invoice, fewer errors, faster approvals and freed-up finance time rather than software savings, and many organisations reach payback within 9 to 12 months depending on volume, exception rate and workflow complexity. Model the payback period against the conservative target, not the optimistic one — a business case that lands early is worth more than one that overpromises.
Questions to ask before signing
- What exactly counts as a billable invoice — do credit notes, duplicates and rejected documents count?
- Are approvers, auditors, suppliers and read-only users billed seats?
- What happens if we exceed the volume band — overage rate, or forced tier upgrade?
- What is the implementation fee, and what specifically does it include?
- Which features sit behind a higher tier: PO matching, multi-entity, multi-currency, supplier portal?
- What are the per-payment fees by method, including FX markup on international transfers?
- Are support, updates and workflow changes included, or chargeable?
- What is the cost of adding an entity, a currency or a second ERP mid-contract?
- What contractual touchless rate or accuracy level underpins the ROI case being presented?
- No per-user or per-approver fees — unlimited users, approval roles, suppliers, entities and currencies. Rules-based approvals can route to whoever genuinely owns the spend.
- Everything included — the full AP automation suite, workflows, audit trails, controls, ERP integration and training, with no add-on modules.
- AI plus human validation — the data service delivers 99%+ accuracy without your team error-correcting extractions.
- No licence fees on the PO module — the purchase order module rolls out across departments without per-user charges.
- Free workflow changes — rule updates and workflow modifications are not chargeable, and supplier statement reconciliation is priced separately on statement volume rather than bundled into a higher tier.
The last one matters most. A quote is only cheap if the automation actually works — a platform that extracts data at 85% accuracy leaves your team correcting one invoice in seven, and that labour never appears in the vendor comparison. There is a fuller argument for treating automation as capacity rather than cost in why AP automation should be your next hire, and practical buying guidance in our AP guides library.
How Kefron AP pricing works
Kefron AP uses a pay-for-what-you-process model: pricing is driven by invoice volume, with a one-off implementation fee based on workflow complexity and ERP environment. There are no per-user or per-approver charges, which removes the trade-off between good approval governance and cost.
More on the reasoning behind that structure is on why finance teams choose Kefron, and if e-invoicing mandates are on your roadmap, e-invoicing compliance is worth factoring into any multi-year pricing comparison.
|
Organisation
|
Result
|
|
Applegreen
|
92% reduction in invoice processing time
|
|
CVS
|
67% reduction in approval time
|
|
200 Degrees Coffee
|
86% of time saved processing month-end
|
|
Emma Bridgewater
|
12-month return on investment
|
Full write-ups for these and others — including Lambeth Council at 7,000 invoices a month, Riviera Travel on Sage 200 and Smith Cameron Group across seven entities — are in our AP automation case studies. You can also watch the platform in action before requesting a quote.
Next step
Full detail on tiers, inclusions and implementation is on the Kefron AP pricing page, or share your invoice volume and workflow details for a tailored proposal within 24 hours.
|
Frequently asked questions
How much does AP automation cost?
There is no single figure, because cost is set by the pricing model rather than a list price. Per-invoice vendors charge a fee per document with volume bands; subscription vendors charge for a defined invoice range; seat-based vendors charge per user. On top of the recurring fee, expect transaction fees on payments, a one-time implementation charge, and higher tiers for features such as PO matching or multi-entity support. The only meaningful comparison is total cost of ownership across the full term against your own current cost per invoice.
What are the common pricing models for AP automation platforms?
Five models are in common use: per-invoice pricing with volume bands, flat subscription for a defined invoice range, per-user or per-seat licensing, transaction-based fees on payment execution, and hybrid models combining a platform fee with per-invoice or per-payment charges. Many vendors apply volume tiers so the unit rate falls as invoice volume rises.
Which AP automation pricing model is cheapest?
It depends on which variable your organisation scales on. Per-invoice suits low or unpredictable volume; flat subscription suits steady high volume; per-user only works for small centralised teams with few approvers. The cheapest headline rate is frequently not the cheapest total cost once transaction fees, implementation and module upgrades are added.
Why is per-user AP automation pricing a problem?
Because it indexes cost to your approval policy rather than your invoice volume. Organisations that route invoices to the budget owner who actually incurred the spend need more approvers, so better financial control costs more. Always check whether approvers, auditors, read-only users and suppliers count as billed seats.
What hidden costs should I expect in AP automation pricing?
Four layers beyond the published rate: transaction fees on payments including FX markup and card interchange, one-time implementation and ERP integration, features gated behind higher tiers such as PO matching or multi-entity support, and the internal labour still needed for exceptions the platform cannot handle.
How do I calculate my cost per invoice?
Divide total AP costs over a period by the number of invoices processed in that period. Include labour and overhead, software, scanning and postage, supplier query handling, late payment charges and reprocessing caused by errors. Ardent Partners puts the average at $9.40 per invoice against $2.78 for best-in-class teams.
How much does AP automation implementation cost?
Implementation is quoted separately from recurring fees and covers process design, ERP integration, data migration, testing and training. It scales with workflow complexity, number of entities and ERP depth rather than invoice volume, so multi-entity and multi-ERP deployments cost more to set up regardless of how many invoices they process.
How do I build a business case for AP automation?
Start with your current cost per invoice including query handling, late payment charges and reprocessing. Set a conservative target using the gap to best-in-class benchmarks. Model total cost of ownership across the full contract term rather than year one, covering platform fees, transaction fees, implementation and module upgrades. Then add the non-financial case — faster month-end close, audit readiness, duplicate payment prevention and capacity released without new headcount.
How quickly does AP automation pay back?
Many organisations reach payback within 9 to 12 months, depending on invoice volume, exception rate and workflow complexity. Return comes from lower cost per invoice, fewer errors, faster approvals and freed-up finance time rather than from software savings alone.
What determines Kefron AP pricing?
Invoice volume processed, following a pay-for-what-you-process model, plus a one-off implementation fee based on workflow complexity and ERP environment. There are no per-user, per-approver, per-supplier or per-entity charges, and support, updates and workflow changes are included.