Activity Based Costing — How Much Does Processing One Invoice Really Cost Your Company

The Answer You Won’t Find on Your P&L

Most CFOs don’t know this number. They know what they spend on the finance department. They know what SAP licenses cost. They know the training budget. But they don’t know what it costs to process one invoice, one purchase order, one complaint. These costs are dissolved into departmental budgets like sugar in tea — they’re there, but invisible to the naked eye.

And yet the benchmarks are clear. The American Productivity & Quality Center (APQC) reports that the cost of processing a single invoice in 2023 ranged from $1.77 at best-in-class companies to $10.89 at the weakest performers. Ardent Partners reports an average of $12.88 for companies without automation. Other sources — Levvel Research, Parseur — cite a range of $12–26 per invoice for manual processes.

Do the math. If your company processes 2,000 invoices a month and each costs $15 to handle, the annual cost of the process alone — not the invoice values, just the handling — is $360,000. For clicking, checking, correcting, and approving.

Activity Based Costing exists so you can see that number. And do something about it.

What ABC Is — and Why Traditional Accounting Won’t Show It

Traditional cost accounting distributes overhead proportionally — by machine hours, headcount, or square footage. This is simple and convenient. The problem is that it assumes every product, customer, and process generates costs in the same proportion as all others. That’s simply not true.

Imagine two products in one factory. Product A is a run of 10,000 units — one machine setup, one quality cycle, one documentation package. Product B is a run of 50 units to spec — separate setup, separate QC, separate documentation, dedicated client communication. Traditional cost accounting will bury the cost of serving Product B in the general departmental budget. ABC isolates it and assigns it to the specific activities that product generates.

Activity Based Costing is a methodology that assigns costs to activities — not departments. Every activity has its own cost driver: the number of machine setups, the number of invoices processed, the number of service visits. The result? You see what it costs to serve a specific customer, a specific product, a specific process. Not approximately — with a precision that, as research by Adetayo et al. (2025) shows, is 10–15% higher than traditional methods.

This is not a new idea. Robert Kaplan and Robin Cooper described ABC in the late 1980s. Thirty years, over 1,200 academic papers. But paradoxically, despite its recognised value, implementing ABC in its classic form was so labour-intensive that many firms abandoned it halfway. Kaplan himself admitted this in the Harvard Business Review in 2004, presenting a simplified version — Time-Driven ABC — which requires estimating just two parameters: the cost per minute of a resource and the time needed to perform a single activity.

The Accounting That Lies by Omission

Here is a scenario we encounter regularly.

A manufacturing company serves 200 customers. Management believes all customers are “roughly equally profitable” — because the P&L shows aggregate revenue and aggregate costs, and the margin looks healthy. Nobody asks how much it costs to serve the customer who orders 50 times in batches of 3, versus the customer who orders once a quarter in batches of 3,000.

ABC answers that question. And the answer can be uncomfortable.

Servicing a small order generates the same administrative activities as a large one: order intake, invoicing, approval, shipping, after-sales. When you assign the costs of those activities to each order individually, you find that 20–30% of customers produce a negative margin once service costs are accounted for. Revenue from them is lower than the cost of serving them. The company is losing money on them — but doesn’t see it, because traditional accounting smears costs across the whole organisation.

This is not a theoretical scenario. It’s a result that repeats across companies of all sizes and industries. Only the scale of the CFO’s surprise varies.

The Cost of Inaction — the Number Nobody Asks About

There is a category of costs that ABC surfaces with surgical precision. We call it the cost of inaction — Cost of Inaction.

It is not the cost of doing something badly. It is the cost of doing nothing. The cost of maintaining a process that “works well enough,” even though it could run three times faster. The cost of an employee who spends 30% of their time correcting errors from manual data entry — which Gartner confirms is the real average in finance departments.

A manually processed invoice costs $12–26 and takes an average of 17.4 days (Ardent Partners). An automatically processed invoice costs $2.50–4 and takes 3–5 days. The difference per invoice: $10–22. On one thousand invoices a month — $120,000 to $264,000 per year. This isn’t “automation savings.” It is the cost the company pays today for the absence of action.

ABC lets you calculate that cost with precision — not at the department level, but at the level of a single activity. How much does one manual approval cost? How much does one call to a supplier asking about a missing PO number cost? How much does correcting one data error cost? Ardent Partners reports that correcting a single error in the invoicing process costs an average of $53, when employee time, system corrections, and payment delays are included.

When you add up these small items — because each one looks trivial in isolation — the total is surprising. Not thousands. Hundreds of thousands.

ABC in the QA10 Architecture — X-Ray Before Surgery

In our glossary we define ABC as a method that precisely assigns expenditure to specific activities, rather than blurring it into overhead. In QA10 practice, Activity Based Costing is the financial diagnostic that precedes every automation decision.

The scheme works as follows. Process mining shows how a process actually runs — how many steps, how many variants, where loops emerge. ABC overlays the cost layer on that map — each step gets its price tag. The result: you know not just that an invoice passes through seven approval stages, but that five of those seven cost the company $8.40 per invoice, and two cost $0.60. You know where to cut.

Only then does RPA enter — automating specific activities, chosen on the basis of hard data rather than a department manager’s intuition. That is the difference between a company that “implements automation” and a company that knows what to automate and why.

On our website you’ll find a cost of inaction calculator — a tool that lets you estimate how much your company is losing on processes that “work well enough.” It’s a simplified model. A full ABC analysis requires access to operational data and time. But the calculator shows the order of magnitude. And that order of magnitude is usually enough to pick up the phone.

Why Companies Don’t Calculate Activity Costs — and Why They Should Start

Traditional ABC implementation required months of work: staff interviews, process mapping, time estimation. Kaplan and Anderson wrote about it directly — companies gave up because the cost of the analysis was approaching the cost of the problem it was meant to solve. In that era, they were right.

Today the situation has reversed. ERP, WMS and CRM systems record timestamps of every operation. Process mining extracts actual times and process variants from those logs automatically. There’s no need to ask people how long it takes them to approve an invoice — just check the logs. Time-Driven ABC, combined with process mining, delivers results in days, not months.

The barrier to entry has fallen. The value of the information has risen — because in a company that processes 200+ documents a month manually, the hidden costs embedded in processes are proportionally higher than in a company that has already automated the basics.