We've spent thirty weeks in this space making the same argument from different angles. Price the problem before you build the solution. The money is in the boring workflows. Take the robot work off your people so they can do the work only people can do.
It's worth ending with the scoreboard — what the pattern actually looks like when you run this method against real businesses, and what it costs.
The projects that cleared
They have a family resemblance, and once you see it you can't unsee it.
They were already happening. Every project that worked was pointed at a workflow the company was already performing, by hand, hundreds of times a month. Nobody had to change a behavior. Nobody had to believe in a new strategy. The work existed, it was expensive, and we replaced how it got done — not whether it got done.
They were measurable before we started. We could say, with a straight face, "this takes 5.2 minutes today, 400 times a month, at a $45 loaded rate." That baseline is what makes a return provable rather than assertable. If we couldn't establish a baseline, we couldn't guarantee anything, and we didn't take the project.
They had exceptions, and we designed for them. Not one successful system handles 100% of cases. The ones that work handle 80-90% cleanly and route the rest to a human with the context assembled. The exception path isn't a failure of the design. It's the design.
Somebody on the inside wanted them. Every project that stuck had a person — not the owner, someone doing the work — who was genuinely relieved it existed. Every project that struggled was one we built for a sponsor and delivered to a team that hadn't been asked.
The projects we killed
We said don't-build more often than we said build. Here's the distribution of why.
Volume wasn't real — somebody was solving for a fear rather than a fact, and when we went and counted, the fact was small. This is the most common one and it costs nothing to discover, as long as you look.
The math didn't clear. A $9,000 problem and a $35,000 build. There's no clever scoping that fixes that. There's just a different project you should be doing instead.
The process was about to change. A platform migration, a regulatory change, a new hire who was going to own it differently. Automating a process that's about to be replaced is the most expensive form of tidiness there is.
The workflow shouldn't have existed. Genuinely our favorite outcome. A report nobody reads. A double-entry created by a merger in 2018. An approval step that traces back to an incident everyone has forgotten. The fix was free, immediate, and made someone's week.
What surprised us
The consequence costs were always bigger than the labor costs. Always. We'd build a careful hours-based model, and then the client would mention, offhandedly, that they lost a $200,000 relationship last year because a package went out late — and the entire economic case would rewrite itself in a sentence. If you're only counting hours, you're underselling your own problem by half.
Nobody was laid off. Not once, in any engagement, at any client. This wasn't a policy we imposed; it's just what happened, because at owner-led scale there is no slack to cut. What we saw instead, repeatedly, was people getting promoted into the work they'd always been capable of — and, more than once, a hire that had been shelved for two years suddenly becoming affordable.
The best projects came from the people doing the work, not from the strategy conversation. The single highest-value question in this entire method is: what would you pay to never do again? It outperforms every framework, every maturity model, and every offsite.
Speed of trust matters more than speed of delivery. The clients who got the most out of this weren't the ones who moved fastest. They were the ones who shipped something small, saw it work, believed it, and then went again. Four small wins in a year beats one ambitious system every time — partly because the small ones ship, and partly because each one buys the credibility to fund the next.
What we still believe, at the end of thirty of these
The market is going to spend the next few years telling you that AI is how you employ fewer people. It's a clean story, it fits on a slide, and it will be wrong for your business specifically — because you don't have a headcount problem. You have a capacity problem, and a payroll that's partially spent on work nobody should be doing.
We use AI and automation to make businesses more human. What that means, in practice, is unglamorous: find the repetitive work, price it honestly, build only what pays for itself twice over, say no to the rest, and hand your people back the hours.
What they do with those hours is where your business actually gets better. They call the client. They train the new person. They catch the thing. They think.
That's not what machines are for. That's what machines are for making room for.
Price one workflow this week. Just one. The number will tell you what to do next — and if it says don't build it, that's a good week too.