How a 73% Reduction in Invoice-Processing Cost Was Counted — Without Replacing Accounts Payable

Finance and operations leaders are right to ask for a return before scaling AI. A pilot is straightforward. Production is not. This article sets out how we build a defensible business case on a representative accounts-payable scenario: high volume, repetitive, and document-heavy. The measure is cost per unit of work, not a general claim of productivity.

Start from the unit of work

Return starts with a measurable unit: cost per invoice processed, cycle time from receipt to posting, the error rate that causes rework, or payment leakage. We baseline those metrics with finance and with the teams doing the work, not only with IT, so savings are tied to labour, penalties, and working capital.

In a typical manual pipeline, people split PDFs, key fields, match purchase orders, chase exceptions by email, and re-enter data in the ERP. Labour cost is direct. The less visible cost is delay, including missed discounts and duplicate payments, and the audit burden when spreadsheets and inboxes are part of the control record.

What an agentic document pipeline changes

The target architecture usually combines optical character recognition or native digital text, extraction with validation rules, retrieval over vendor history and contracts for edge cases, exception queues with suggested actions, and straight-through posting when confidence thresholds are met. People move from data entry to true exceptions and supplier relationships.

Building the ROI model

We model annual invoice volume, minutes saved per document at each stage (classification, line extraction, matching, and coding), loaded labour rates, and the reduction in rework tickets. We include implementation cost: integration, document storage, monitoring, and ongoing model operations or prompt tuning where they apply. We then apply a risk buffer for adoption. Not every process reaches straight-through processing on day one.

• Benefit: volume × time saved × cost of time + avoided penalties / duplicate payments (where measurable).
• Cost: build, licensing, infrastructure, security review, training, and run-state support.
• Payback: simple payback horizon and sensitivity analysis on volume and adoption.

Where a large team still processes documents manually, a material reduction in handle time per document can become cost avoidance. That saving usually arrives in phases, as exceptions fall and confidence in the automation rises. The 73% figure in this work is a cost reduction: a composite of labour redeployment, fewer rework loops, and supplier consolidation after controls improved. It is not a reduction in processing time alone, and it was not immediate.

What weakens the business case, and how we address it

• No owner for exceptions and continuous improvement after go-live.
• Weak integration: the model sits beside the process while people still re-enter data in the ERP.
• Skipping evaluation sets for extraction quality across suppliers and languages.
• Ignoring audit: logs, retention, and access control must match finance policy.

If you are preparing a case for the board, we help quantify baselines, define success metrics, and align IT, finance, and operations on a phased plan, so the figures presented match the operation that will run.