The math nobody does
Take a common example: invoice processing. A mid-sized company receives 200 invoices per month. Each invoice requires someone to open the email or scan, extract the relevant data (vendor, amount, line items, PO number), enter it into the accounting system, match it against a purchase order, route it for approval, and follow up on exceptions.
That process takes 15-25 minutes per invoice. At 200 invoices per month, that's 50-83 hours of human labor. At a blended cost of $40/hour (including benefits and overhead), you're spending $24,000-$40,000 per year on a single workflow that an AI agent handles in under 2 minutes per invoice.
And invoices are just one workflow. Stack the same math across support ticket triage, lead qualification, expense reporting, compliance checks, and employee onboarding, and most organizations discover they're spending $200,000-$500,000 per year on work that should be automated.
Why the cost stays hidden
Three reasons:
The work is distributed. Nobody owns "invoice processing" as their job title. It's spread across AP clerks, office managers, and whoever happens to be available. Because the cost is spread across multiple people and departments, it never shows up as a single line item.
There's no baseline to compare against. If you've always spent 70 hours per month on invoice processing, you don't think of it as a cost — you think of it as "the way things work." Without a benchmark for what automated processing looks like, the manual cost is invisible.
The secondary costs are uncounted. The 15 minutes per invoice is just the direct labor. It doesn't include the cost of errors (duplicate payments, missed early payment discounts, vendor disputes), delays (approvals stuck in inboxes), or opportunity cost (the AP clerk who could be doing vendor negotiations is doing data entry instead).
How to make the cost visible
Run this exercise with your team this week. Pick your top 5 highest-volume manual workflows and calculate: monthly volume, time per instance, monthly hours, hourly cost, and monthly cost.
A conservative example for a mid-sized operation: five common workflows totaling $8,685/month — $104,220/year.
The automation benchmark
At Fangre, we typically see 70-85% automation rates on these workflows within the first 30 days of agent deployment. At 75% automation on a $104K annual spend, the net annual savings after agent costs is over $66,000 — a 6.5x ROI.
And that's before counting the throughput gains, the error reduction, and the speed improvement.
What to do with this
The exercise above takes 30 minutes. The output is a one-page business case that speaks your CFO's language. Don't pitch "AI automation" — pitch "we're spending $104K/year on work that costs $12K/year automated."