TL;DR: Rory Sutherland's Alchemy is a useful provocation for Lean practitioners. Lean already knows how to test small, low-cost ideas without demanding an airtight ROI in advance. We sometimes forget.
A Small Idea, Quietly Killed
A few years ago, I was in a hospital workshop when a nurse mentioned she had an idea for a small change to her unit's supply layout. Her manager, who had been nodding along all morning about Kaizen, asked what the projected savings would be. The nurse said she didn't have a number. Could she put one together? Sure, she said. She'd let her know.
I never heard if the idea got tried. I'd be willing to bet not.
I have watched some version of that conversation play out many, many times. The polite pause. The reasonable-sounding question. The quiet calculation, on the part of the person with the idea, that the cost of pursuing it just exceeded the cost of letting it go. Nobody in the room would say they were against improvement. Most of them would have told you they were for it. The idea died anyway.
That is the pain I want to sit with for a minute. Improvement does not usually die from outright refusal. It dies from being asked, gently and repeatedly, to prove itself before it gets a chance to exist. The questions sound responsible. Where is the data? What is the ROI? Has another hospital published this? Why exactly would this work?
For a hard-to-reverse decision involving real money or real risk, those are the right questions. Applied to a six-foot supply cart move or a bottle of bubbles in an exam room, they are how a Kaizen culture quietly suffocates.
What Sutherland Is Actually Saying
I have been listening to Rory Sutherland's Alchemy. His argument, as I read it, is that human behavior often does not follow the logic that economists or engineers (like me) expect it to. People respond to context, framing, status, reassurance, meaning. Some interventions sound silly and work anyway. The value gets discovered empirically, not deduced from first principles.
Sutherland likes data well enough. His complaint is about monoculture. When one kind of logic – the kind that demands a clean rational explanation before a small test – gets treated as the only legitimate way to evaluate an idea, the rest of human behavior falls outside the conversation. And a lot of human behavior is where the gains are.
For Lean people, I think this is a healthy provocation. We tell ourselves we love small tests of change. Sometimes we love them only when the explanation is tidy.
Lean Should Already Be Compatible With This
PDSA is built for this. Kaizen is built for this. The whole point of testing on a small scale is that you do not need to know in advance why something will work. You need to know it is small, reversible, and worth learning from.
Jeff Liker once told me that Toyota, when faced with a new idea, is more likely to say, “Please try it, and you will learn,” than to debate it. That is not a casual cultural detail. It is a method for getting more learning out of a system per unit of effort.
Treating every idea like it needs the same scrutiny as a capital project is one of the easier ways to kill a Kaizen culture without realizing you are doing it. The big change with hard-to-reverse consequences deserves rigorous evaluation. The change you can try Tuesday and undo Wednesday does not.
(Jeff Bezos at Amazon also referred to the difference as one-way doors and two-way doors).
Deming Was Always Talking About Psychology
People often label Deming a statistician. He certainly cared about variation, and his work changed how I think about charts and metrics. But his System of Profound Knowledge has four parts, and one of them is psychology. Alongside variation, systems, and knowledge.
Gregg Stocker told me on the podcast that if he had to name one Deming idea he wishes more leaders understood, it would be psychology. Too many leaders, he said, get promoted on technical capability and never build fluency in how people actually work. I have seen that pattern in action. A leader who can read a Process Behavior Chart but goes silent when a frontline staff member says, “I'm afraid to bring this up.”
A psychologically informed Lean practice can still be rigorous. It tends to be more realistic about how change actually happens.
Obliquity, and the Trouble With Reaching Directly
I have written before about John Kay's book Obliquity. Kay's argument is that some goals are best pursued indirectly. Profit is one of his examples. Companies that obsess over quarterly profit often produce less of it than companies that focus on customers, quality, and the people doing the work.
I borrowed the same idea in Lean Hospitals when discussing cost. Hospitals that lead with cost-cutting tend to underperform on cost over time. Hospitals that focus on waste, quality, and flow tend to see costs come down as a result.
Psychological safety works the same way. You cannot command people to feel safe. You cannot set a target for psychological safety scores and bully your way to a higher number. Try it and you'll get gamed survey responses long before you get a healthier culture. Safety emerges when leaders respond constructively to bad news, when mistakes are studied instead of punished, when problems get surfaced instead of hidden.
The metric is what the work produces. Reaching directly for the metric, instead of for the work that produces it, is what gets leaders into trouble.
“We Don't Know Yet”
A leader asks why a small change worked. The team does not have a complete answer. Eyes drop. Someone clears their throat.
That moment of embarrassment is worth examining. In a learning organization, “We don't know yet” should be allowed, celebrated, and rewarded – as long as it is followed by curiosity. Sometimes practice precedes theory. Sometimes you observe that something works before you understand the mechanism. That is how a lot of real-world discovery actually happens.
Sutherland makes this point in the language of behavioral science. Lean ought to make it in the language of empiricism. The worst response to “we don't know yet” is to make people feel foolish for not having the answer ready in the first place.
Bubbles for Babies
The example that keeps coming back to me is one Joe Swartz and I wrote about in Healthcare Kaizen. Hope Woodard, an ultrasound tech at Franciscan St. Francis Health, had a problem she watched every day. Young children were uncomfortable during ultrasound procedures. The probe was cold and hard. The kids would not stay still. Parents got frustrated. The techs got frustrated. Image quality suffered. Everyone had a worse day.
If you have ever tried to keep a small child still while you press something cold against them, you know it does not respond well to a logical argument.
Hope brought in a small bottle of “wedding bubbles” she had at home and asked parents to gently blow them over their child during the procedure. The children went still. Their faces softened. The images got sharper. Staff and parents had a less stressful afternoon. The estimated cost savings line on her Kaizen card read: “Priceless!”
This is a near-perfect example of what Sutherland is talking about. The intervention sounds too simple. There was no ROI calculation. There was a reasonable hypothesis – children get distracted by bubbles – but no certainty about whether it would work in this specific clinical setting. She tested it. It worked. The team learned something.
If she had been required to prove in advance that bubbles would calm children during ultrasounds, she would not have run the test. She probably would not have brought it up.
One Standard Doesn't Fit Every Situation
I want to be careful with the bubbles story, because it can get misread.
Standard work does not mean “blow bubbles for every baby.” Some children will not respond. Some parents will not want them. Some children will be soothed more by music, a familiar toy, or a clear, calm explanation of what is about to happen. The opposite of a good idea can sometimes be another good idea, as Sutherland likes to put it.
A useful standard might be: assess what this child and parent need; offer a small set of safe calming options; try one; adjust based on response; share what worked. The standard is the method of learning. The tactic varies.
Discharge: Where the System Isn't Ready
Healthcare has plenty of places where a Sutherland-flavored experiment could earn its keep. Discharge is one I think about often.
Most of us have been on the family side of this. Your loved one has been told they can go home. The nurse smiled. The doctor signed off. And then you wait. An hour passes. Two hours. You ask politely. Someone says they're working on it. The patient sits in the chair, dressed, growing tired. The room next door is empty and waiting for the next admission.
The patient is ready. The system is not. Pharmacy is behind. Transportation is delayed. Paperwork sits unsigned. A handoff between teams missed a step. Nobody is doing anything wrong. Everyone is doing their part of a coordination problem that no one person can see.
The interesting question, in Sutherland's spirit, is whether some kind of visible signal could change the social dynamics of discharge coordination. A simple “ready to go home” indicator outside the door. A shared countdown. A visual cue that says, in effect, “this person is waiting on us, together.”
I will not tell you that works. I have seen ideas like this energize teams, and I have seen them stress staff out. That uncertainty is exactly why it is a candidate for a small test, with input from the people doing the work, clear stop conditions, and someone paying attention to whether the test helps or harms.
The answer is not “install countdown clocks everywhere.” The question is what small signal might shift the social dynamics of this work, and how do we learn from trying.
What Holds the Odd Ideas Up
Teams do not propose unconventional ideas in environments where unconventional ideas get mocked. They learn what is allowed by watching what happens to the early ones.
When small experiments are welcomed, when failed tests are treated as learning, when leaders respond with curiosity instead of an arched eyebrow, more ideas come forward. Some of those ideas will sound a little odd. A few of those odd ones will turn out to matter most.
This is why I am cautious about leaders who say they want a culture of innovation but cannot tolerate a small idea that arrives without a business case. Innovation comes from people who feel safe enough to try things they cannot yet justify. Departments named “innovation” rarely produce it.
What This Isn't
I do not want this read as a license to skip evidence-based medicine, ignore patient consent, work around infection control, or substitute vibes for data when the stakes are high.
A test worth running has some reasonable criteria. It is low cost. It is low risk. It is reversible. It is respectful to patients and staff. It is measurable enough to learn from. It was designed with input from the people doing the work. It can be stopped quickly if harm or burden shows up.
Most ideas I have seen die in healthcare meet all of those criteria and never get tried, because someone in the room asked for an ROI calculation no one could produce.
A Better Question to Ask
Maybe the better question is not “can we prove this will work before trying it.” Maybe it is closer to “is this safe, small, cheap, reversible, and worth learning from?”
I think about that nurse from the workshop sometimes. I do not know what her idea was. I do not know whether the supply cart ever got moved. What stays with me is the small, almost invisible moment when an idea got priced out of existence by a question that sounded reasonable.
What small thing have we talked ourselves out of trying because we could not explain in advance why it would work?
