Back in Chapter 13, I made this argument as philosophy. This is the practice version, and the practice version is a habit you have to catch in the act.
Because you don’t design for the idealized human once, in some grand architectural decision you could locate and reverse. You do it in small strokes, constantly, every time you build anything.
—
You do it when you write an SOP and assume it will be read, absorbed, and recalled under pressure — by someone who will actually meet it mid-crisis, at 4:50pm, three tabs deep.
You do it when you buy a tool and assume adoption, because the demo went well and the demo was attended by people at their most rested and least interrupted.
You do it when you design a handoff that works if everyone communicates proactively — which is to say, a handoff that works if nobody ever needs a handoff.
You do it when you set a process that requires sustained discipline from people whose calendars you have seen. When onboarding assumes context transfers by proximity. When the meeting assumes everyone processes information live, out loud, at speaking pace — and treats the person who thinks best in silence, two hours later, as disengaged.
Every one of these designs is a small bet on the same fictional employee. The one who always follows the process. Who never needs context repeated. Who reads everything, retains everything, and never has a Tuesday where their executive function stayed home. Who doesn’t bring their nervous system to work.
That person does not exist. That person has never worked for you. And you have never once hired them, though every process you’ve ever drawn assumed they’d show up eventually.
—
So here’s the practice. One question, asked at design time, every time:
*Who does this break for, on their worst plausible day?*
Not their catastrophic day. Their ordinary-bad day — distracted, overloaded, mid-flare, post-daycare-dropoff, running on the sleep they actually got. If the process only works above a certain threshold of human condition, the process has a failure rate you’ve been booking as a personnel issue.
Design review used to mean: does this work when everything goes right? Reality review asks: what happens to this when a real person meets it on a real day? Does the system carry the memory, or does the human? Does the truth-telling step cost status? Does the critical handoff survive the one week its owner is out?
Ask it and the fixes are rarely expensive. Checklists at the point of work instead of manuals on a drive. Defaults instead of discipline. Context that travels with the task. One owner per seam. The system remembering so nobody has to.
What’s expensive is the current setup — you’re just paying for it in a currency nobody audits: workarounds, quiet burnout, revenue leaks booked as coaching problems, and the slow exit of your best people, who were never bad at the job. They were good at a job that was designed for someone else.
—
The organizations that scale well from here will not be the ones that finally found idealized humans. There aren’t any. There’s no shipment coming.
They’ll be the ones that stopped waiting — and built for the people who actually show up.