A few weeks back, I wrote about how you’d followed the advice, sorted out your processes, and you still were left with a system that didn’t quite work. I talked about how off-the-shelf standard systems can now be flexed to fit your unique shape because of how AI can build for you. This newsletter is about a different layer of the same frustration.
The two jobs of a system
Most founders think of a system as having one job, to move work through, let people know what’s next, let people know if they’re on track, see what’s in the to-do pile.
But a system should be doing another job, too. It should capture the underlying economics of the business. Capturing what a unit of value actually looks like. What’s the margin on each project, each client? What is each role actually worth?
It should capture data so it can answer a bunch of questions for you:
Are we pricing staff cost right?
Are we attributing shared cost right?
What’s the return on each role?
How this happens
Many systems can’t answer these questions simply because they didn’t know they were supposed to.
You iterate the process against what the client values. Of course you do. That’s where the attention had to go. You work through optimising how work gets done, tweaking to make sure you were giving the client the best of what they’re paying for. While that was happening, the economic value of the business was shifting underneath, and you didn’t notice because you couldn’t have. You weren’t supposed to be looking.
A founder I spoke to recently had been doing all the sensible things. He’d worked out his overheads, factored in that nobody is productive 100% of the time, applied a margin, and arrived at a costing number - which then felt impossibly high. He’d split his overheads down the middle between the two sides of his business because that’s how the expenses split. But the two sides made money in completely different ways, and the costing structure couldn’t see that. The number wasn’t wrong. The shape underneath it was.
This is why so many founders end up asking ChatGPT what their hourly rate should be. Not because they think it knows. Because their own system doesn’t.
The system was built to carry the work, and it hadn’t thought to know it needed to carry and capture the other data points you’d need. It was missing half its job description and nobody had noticed.
This is how it had to be
You couldn’t have seen it from where you were standing. How the business behaves economically only becomes apparent after you’ve evolved it. Until you’ve done enough laps around the block, you don’t know what will matter. Your business reveals itself through iteration and not before.
The moment to pause
Once you get to the stage where the iteration stops and you’ve landed on how you do it, the system can become a deliberate design and begin fulfilling that second half of its job description. That’s the point at which you can pause and redesign the system around the data that answers those real questions.
If your system handles the work but can’t tell you whether it’s making you money, project by project, client by client, role by role, that’s the half it was never asked to do. And it’s a fixable thing, once you know to look.
If you’d like to talk through it, I run free 15-minute fit calls - there’s a link in my featured section. You tell me what’s going on, I’ll tell you straight whether I can help.