The Only Cost I Want to Go Up

I was on a call this morning, talking through a team member's scorecard.

We got to the line item for API costs, and I said something that confused him.

"I want this number to go up."

He paused.

Costs going up?

That's backwards. Every business owner knows you're supposed to CUT costs, not celebrate them.

But…..$800 in API costs replaces $80,000 in labor.

Read that again.

APIs—application programming interfaces—are how software talks to other software.

When you build an automation that pulls data from one tool and pushes it to another, or when you use AI to process a document, you're making API calls.

And APIs are USAGE-based. This matters more than you think.

You can waste OODLES of money on software.

Buy a bunch of Salesforce licenses nobody uses.

Sign up for project management tools that collect dust.

Rack up subscriptions to things that sounded good in a demo.

Seat-based software is a liability—you're paying whether anyone touches it or not.

API costs don't work that way.

It is almost impossible to waste money on them unless you literally set up a fake automation that runs in circles.

The meter only ticks when work actually happens. When automations fire. When bots do the tedious stuff your team used to do.

Every dollar spent on API calls means work is getting done—work that used to require a person sitting at a desk, clicking through screens, copying and pasting, scheduling things manually.

This is PARTICULARLY true if you run a service business. Your margins live and die by labor costs. Every hour someone spends on repetitive work is an hour you're paying for. In a SaaS business, maybe the math is different—I honestly don't know.

But in services, the equation is brutal and simple: people cost money, and most of what they do is routine.

So I told him: your job is to find more places to insert API usage into our workflows. Candidate scheduling? Build a bot. Data entry? Automate it. Every piece of manual work you can hand off to a machine shows up in that number.

And here's how we know it's working: revenue per employee goes up. If we're doing more business with fewer people because the machines are handling the boring stuff, that's the win. The API cost is the leading indicator. Revenue per employee is the trailing one.

It's counterintuitive to have a KPI where you want costs to increase. But $10,000 in API costs is hard to generate. You have to actually build something useful, integrate it properly, make it work. And that $10,000 probably replaces $100,000 in labor. Maybe more.

So yeah. Spend my money on APIs. You literally cannot spend it fast enough.

Yallah Habibi,

Jon

P.S.

Sagan gets good reviews: