According to Ardent Partners' "AP Metrics That Matter" 2025 report, the average company spends $9.40 to process a single invoice. Best-in-class automated operations get that down to $2.78. That's a 3.4x gap, and it's been sitting there for years.
So why did the IFOL 2025 Report find that 66% of finance teams are still manually keying invoices into their ERP? Not a legacy stat. That number actually increased year over year.
Because the tools aren't the problem, it's generally the implementation.
We've watched companies buy AP automation, demo it for leadership, and quietly abandon it six months later. The AP team ends up doing more work than before, not less. The OCR looked great in the sales demo. Approval routing made sense on the whiteboard. Then real invoices showed up from real vendors and everything fell apart in ways nobody warned them about.
The capture isn't where things go wrong
Every AP automation vendor leads with capture. Drop a PDF, watch the OCR read it, marvel at the magic. Capture has gotten pretty good, with vendors now routinely claiming 90%+ accuracy on structured invoices. But capture is only the first step, and it's not where teams actually get stuck.
We hear this constantly from implementation teams: the issues don't tend to show up at capture. They show up once invoices move through approvals and get posted back into NetSuite.
That tracks with what we keep seeing in implementations. The tool looks solid at invoice capture, then approvals get disconnected from coding, sync isn't clean, and finance spends the last three days of every month fixing entries before close. What kills your close is the post capture bottleneck, not the OCR.
The inventory item trap nobody warns you about
When you buy inventory and do things properly in NetSuite, there's a three-step flow:
- Purchase Order
- Item Receipt
- Vendor Bill
The item receipt debits Inventory Asset and credits a temporary account called Accrued Purchases (sometimes called Inventory Received Not Billed). When the vendor bill comes in and gets linked back to that PO it clears the accrual. The trail is intact, reconcilable, and auditable.
Here's what happens when your AP automation tool creates a standalone bill for an inventory item without linking it to the PO line.
The system debits Inventory Asset and credits Accounts Payable directly. Accrued Purchases never gets touched. You break NetSuite's native three-way match, bypass Accrued Purchases clearing, and leave your controller reconciling by hand on every close.
This matters most when you're using Advanced Receiving or a separate receipt-and-bill flow, which is most mid market and above NetSuite implementations. If you use Inventory Items on POs, the AP solution must create Bills linked back to the PO line. Standalone Bills with Inventory Items will affect your inventory valuation, and AP vendors don't warn you about this during evaluation.
If you carry inventory, this is the first question you ask any AP vendor. Make them prove it works before you sign. Most entry-level tools can't handle this properly. They'll create standalone bills and your Accrued Purchases account will quietly drift until someone catches it at audit. We've seen this show up at companies running Shopify or BigCommerce storefronts where high order volumes generate hundreds of POs a week, and the AP tool just bypasses the matching entirely.
The inventory item trap
What actually posts to the ledger when a bill is — and isn’t — linked to its purchase order. Pick a posting mode, then click through the five POs and watch the accounts fill in.
Inventory Asset
Asset · DrAccrued Purchases
Liability · CrAccounts Payable
Liability · CrSpikes at receipt, snaps back to $0 when the linked bill posts. A healthy clearing account.
Amounts illustrative. Account names as in NetSuite: Accrued Purchases = Inventory Received Not Billed.
NetSuite Bill Capture: free-ish, with asterisks
Oracle's native Bill Capture module gets the most polarized reviews of any tool in the space. Some teams love it, others... less so.
Bill Capture works fine if you process a moderate volume of invoices, mostly from the same vendors, in US dollars. It learns vendor patterns over time. The OCR can read a sideways phone photo. It connects to NetSuite's three-way match natively, which solves the inventory problem above. And as a native NetSuite capability, there's no separate vendor relationship to manage.
The limitations are real though. Bill Capture support is still limited and region-dependent, with Oracle's documentation and partner guidance not always lining up cleanly across markets and configurations. It was originally US-only; the 2025.1 release extended support to UK tenants using SuiteTax, but availability can vary by account setup.
Multi-language capabilities are thin. There's a 30-page PDF limit per invoice and a one-bill-per-file restriction. The review page isn't customizable. Oracle did add duplicate detection and bulk management features in 2025.1, which addresses some of the more common complaints from earlier versions. But if you're processing invoices in languages other than English or operating outside those supported markets, your options may be limited depending on your specific configuration.
Then there's the approval workflow.
We've heard multiple teams describe it as "really rigid." One accounting manager we worked with found that Bill Capture ended up more expensive than third-party software because every approver needed a NetSuite user license. That complaint comes up a lot, but it's worth digging into.
Lower-cost employee-style licenses may be sufficient for approvals if you configure the role with the right permissions. Most teams don't discover this during evaluation because nobody tells them, and the actual cost depends on your Oracle agreement.
Bill Capture is a reasonable starting point for US-based teams with straightforward AP needs. It's not going to work for multi-entity, multi-currency, or high-volume operations. Know which one you are before committing.
Ramp is great until it isn't
Ramp has a genuinely enthusiastic user base in the NetSuite ecosystem right now. We've heard teams describe switching from Bill.com as "like going from a tricycle to a Ferrari." One controller called it "the simplest integration I ever deployed." Another told us it "cut our AP posting and approval workload by about half."
For basic AP, that makes sense. The interface is clean, sync works well for standard transactions, and for teams processing straightforward contractor invoices at moderate volume, it's genuinely good.
But if your AP involves inventory, purchase orders, or anything beyond simple expense coding, there's a list of limitations that won't come up in the demo.
Three-way matching requires the Plus subscription. And Ramp relies on NetSuite as the system of record for inventory, so all inventory-related POs and item receipts have to originate in NetSuite.
We've also run into issues with decimal quantities not being supported in line items, the PO dropdown only showing NetSuite internal IDs rather than custom PO numbers, and vendors who bill with a deposit-plus-net30 structure that Ramp can't split. Individual issues get patched over time, but the underlying design bias doesn't move: Ramp's AP features were built for expense-style bills, not PO-heavy procurement.
None of this makes Ramp bad. It makes it a bad fit for certain workflows. What we see repeatedly is teams adopt Ramp for corporate cards, discover the AP features, get excited, and start pushing PO-heavy processes through a tool that wasn't built for them. By the time they figure it out they've already migrated 400 vendors over.
"Touchless" means different things to different vendors
Most AP vendors today claim some version of touchless processing. Meanwhile, only 32.6% of invoices actually get processed without human intervention, according to Ardent Partners. Which means the average "automated" AP function is still reviewing two-thirds of its invoices by hand.
Part of the problem is that "touchless" doesn't have a consistent definition. For Charted (formerly SquareWorks), touchless means auto-creating bills when a PO matches exactly or a recurring vendor's pattern is high-confidence. The bill gets created and submitted for approval without anyone reviewing it. For Zone & Co, it means their OCR captures 90%+ accurately, but a human still reviews before posting. For industry analysts, it means fully posted without any human involvement at any stage.
Those are three very different things. When a vendor tells you their tool delivers touchless processing, pin them down. Does the bill get created automatically? Coded automatically? Approved automatically? Posted to NetSuite automatically? Each of those is a separate step, and most "touchless" claims break down somewhere in the middle.
What to actually ask before you buy
Teams that succeed with AP automation ask different questions during evaluation. Failed rollouts almost always spent the demo asking about OCR accuracy. Successful ones asked what happens after capture: approvals, coding, sync failures, month-end reconciliation.
- If you carry inventory, the question is simple: can your tool create vendor bills linked to PO lines for inventory items? Not standalone bills. Linked bills. Ask them to demo it with your actual item types.
- Multi-subsidiary gets complicated fast. How does the tool handle inter-entity vendor bills? Can it post to the correct subsidiary and manage intercompany clearing? This is where integration architecture matters more than the tool itself. Orders that originate in Salesforce or HubSpot and flow through to fulfillment costs in NetSuite need to land in the right subsidiary's AP. Most AP tools assume a single entity.
International payments are their own problem. Does the tool handle FX conversion or push that to your bank? Realized gains and losses need to get captured in NetSuite somehow. Platforms like Airwallex and PayPal add reconciliation complexity that domestic focused AP tools tend to punt on entirely.
And the question everyone should ask regardless of their setup: what happens when sync fails? Because it will fail. Does the tool log the failure? Alert you the same day? Or do you discover it during close?
If you're running Celigo or Workato as middleware, you also need to understand where AP data flows intersect with other sync jobs. A vendor bill that arrives through one integration path and gets modified by another is how you end up with ghost entries at month-end.
It's an integration problem, not a tool problem
We recently spoke with a mid-size manufacturer whose AP process looked like this: invoices hit a shared inbox as PDF attachments, someone opens each one, types header info into NetSuite, matches to PO manually, routes for approval via email, chases down approvers when they ignore it.
They process 1,500 to 2,000 invoices a month. Tried NetSuite's built-in OCR. It choked on half their invoices because every supplier formats theirs differently.
Their problem isn't that good AP tools don't exist. It's that nobody mapped the full flow from invoice receipt to bank reconciliation before picking one. The tool that reads the invoice is maybe 20% of the workflow. The other 80% is how that data flows through approvals, posts to the right accounts and subsidiaries, clears the right accruals, and reconciles at month-end. Picking a tool without designing that upstream flow doesn't fix an AP process; it locks the broken one in.
Want a straight answer on whether any of this applies to your setup? Book a 30-minute call with our team.
