Continuous improvement - Improve Something Today (Page 4)

Some things (like pancakes) you should finish once you start

Would you rather have five half-cooked pancakes or two good ones?

Do this with me:

Griddle up a pancake on one side.
Do not flip it. Instead, remove from heat and set on a plate.
Immediately start the first half of another pancake.
Continue until you have a stack of five half-cooked pancakes. They should be crispy and fluffy on one side, oozing and dripping raw batter on the other.

Oops. We didn't make a meal—we made a disaster.

The 10x pancake-maker

I’ve met busy, expert-level cooks who spend a great deal of time making only the first sides of pancakes. Everyone is hungry, the food goes to waste, but you gotta see how fast they move. Their bias towards action. Velocity. Look at them: a “flow state.” There is always something cooking, so they must be great at this. And the incredible variety of pancakes their menu offers.

What remarkable flavors their breakfasts must have, if they ever manage to finish a single god damned flapjack.

The best pancake

The best pancake is cooked perfectly, ready for someone to enjoy.
A strongly OK pancake might be a bowl of batter next to a hot griddle.
The very worst is a pancake-in-progress, one that's started cooking, and then a doorbell rings a few rooms away. It’s easily interruptible, delicate, a liability if the mind or body wanders.

Started but not finished

Put on your continuous improvement goggles, go look for work that has been started but not finished. You’ll find it everywhere. Gross. Let it be as much cause for concern as that stack of half-cooked pancakes.

Given a pancake-in-progress, you can guess how far along it is. Pour, one side, flip, the other, serve. You can even check the batter and guess how many more you can cook.
But given a basket of nuanced, exploratory projects, everyone believes & reports each activity to be 85% complete… but the reality can be obscure. Plus, everybody is too busy starting all these other projects to quite finish any of them.

It’s enough to put you off your breakfast.

August 5, 2024: Edited for length.

A lousy time to be a good idea

In which “we send our hypotheses ahead, an expendable army, and watch them fall.”

In 1992, Oxford English prof Andrew Nuttall wrote the following:

The human capacity to think provisionally, to do thought experiments, to form hypotheses, to imagine what may happen before it happens, is fundamental to our nature and to our spectacular biological success ... The cleverest thing Sir Karl Popper ever said was his remark that our hypotheses ‘die in our stead’. The human race has found a way, if not to abolish, then to defer and diminish the Darwinian treadmill of death. We send our hypotheses ahead, an expendable army, and watch them fall.

That’s the tidiest explanation I’ve seen of our infinite appetites for catastrophe in media and in the imagination.

There’s much to gain from the activity of worrying about responses to some imagined, awful event. Mental rehearsal prepares us. We imagine. We plan. We ruminate. And as the expendable armies fall, over and over, the mind is cluttered. All this automatically, unwillingly, exhaustively.

We might also stop and turn the armies back and inhabit the silence.

(But that’s for another day.)

This is hard

We’re born experts at doing thought experiments automatically, intuitively, and individually.
But we’re terrible at doing thought experiments deliberately, or being able to speak or write about them.
Even worse, we believe we’re equally great at both kinds of thinking.

The bad news

Think of the containers we’re meant to pour our great ideas into:

Suggestion boxes where ideas go not merely to die but to be forgotten “even by God.”
Big-design-up-front schemes that admit only a single hypothesis. Hope you picked the right idea at the outset of that multiyear effort.
Asinine standards for projects like 6x return on investment, which serve only preclude experimentation and prohibit critical thinking.
Initiatives where everyone is discouraged by an early failure and gives up, with the cry of: “we tried that and it didn’t work!”

These containers are bad, and you can find them everywhere.

The good news

The good news is: there are better ways of working and deciding, together.

Instead of suggestion boxes, create listening systems.
Do set-based design and allow for new trade-offs and discoveries without blowing the budget.
Validate ideas through rigorous PDSA, with small, reversible changes.
Celebrate failure as the thing one can learn from.

All those continuous improvement methods, all the great facilitation approaches, the techniques for self-organization? That’s what they’re for. They’re here, they’re tested, and they’re effective—and they are not easy.

From where I sit, this is the job. Let’s get to it.

October 7, 2024: Edited for length.

How I learned to speak about changes when 19 out of 20 people would rather I shut the hell up

Nobody wants to hear it from the consultant.

Imagine this. A big, important change is happening: reorg, strategic shift, start or stoppage of significant work.

How do folks across the organization want to learn about these changes? Here’s the breakdown from the change management boffins at Prosci:

73% of people surveyed want to hear about the scope and importance of big changes from their executive leadership.
67% want to hear about the impact of changes from their direct supervisor.
And only 4% want to hear ANYTHING from the change management team (hey, that’s me!).

When I first saw this data, my reaction to that 4% figure was a big old “eep” as I thought back to the times I’d communicated change messages myself. Although it wasn’t something I did often, it was clearly something I should do… never.

Instead, it was time to get good at creating space for the right people to send the right messages. To learn how they talk and give them their talking points. To schedule the sessions, invite the people, buy the food, and get the heck out of the way. This is where I learned the importance of visible, participatory, energetic sponsorship for change efforts. Sponsors need to show up, over and over, and speak to the change and why it’s happening now.

This is one of a million (approx) things I’ve learned from Prosci, the outfit behind the figures I mentioned above. Prosci researches change management, then pours what they learn directly into a container of excellent, thoughtful change management tools + methods.

I now believe that, in the same way that every project can be a lean project, every project can be a change management project. None too big, none too small. Maybe the strategy is simple, but it’s still worth thinking it through.

And if you ever find a project where the sponsor won’t speak up, RUN! Life’s too short to try to talk one’s way out of that dismal 4% figure.

TL;DR:

Communications about big changes should come from the right people:
Those in exec/leadership roles…
…And the managers of everybody affected by the change.
Prosci has super great methods to help.
Don’t waste time on projects without active sponsorship.

3-step method for learning anything

“See one, do one, teach one.” ← That’s the method.

First: See one

Watch how the task is done. Have somebody who knows what they’re doing show you how. Take notes. Ask questions. Notice the materials, tools, techniques used.

Second: Do one

Get some first-hand experience. Bring your expert along. They’ll speak up with encouragement or suggestions. Take your time and do it right. Figure out what makes it complicated—there’s always some damn thing.

Third: Teach one

Show somebody else how to do the task. You know how, and you know all the little things they’ll need to know or have on hand to get the job done. They’ll learn just by seeing you work.

You’re done!

This learning model originated 130 years ago, usually attributed to William Stewart Halsted’s surgery dept. at Johns Hopkins University. ‘See one, do one, teach one’ is still used in medical training. It works. In more recent decades, because we can’t have med students slinging a knife and ‘doing one’ on patients anymore, the model has been updated a bit. What happens in medicine today is more like ‘see many, do many with supervision, teach many.’ In other words, they’ve slowed the learning method down for patient safety, but not changed the progression in any basic way.

For knowledge work

Guess what? ‘See one, do one, teach one’ is a great fit for knowledge and service work, because there is so much tacit knowledge embedded within… all that little, vitally important stuff like (these are real examples):

It won’t let you pull the report unless you sort data by these headings, for some reason.

Even though form A is first, it’s easier to start with form C since that asks the same questions in a better order.

The Spanish language brochures are out of date, make sure you update this part first if you hand them out.

The position of the ‘save’ and ‘cancel’ buttons are reversed on this screen.

Try ‘see one, do one, teach one’ sometime. It’s so much better than sending everyone to training (boring) or throwing a grossly outdated playbook at new hires (inhumane, although you should still write a playbook and update it).

Apprenticeship is the method, and one’s peers are the experts. The trainer’s role then becomes that of matchmaker—finding a person ready to teach for each person ready to learn—and of opening a little space and time for learning to arise.

TL;DR:

See one, do one, teach one is a method to teach and learn complicated tasks.
It matches how people learn (by doing)…
…and transmits what’s important for doing the work (tacit knowledge).
You don’t need an elaborate training function to do it.

January 28, 2026: updated formatting & photo.

Dust off those awful process maps. Add these 3 things to them to get people thinking

Get those process maps off the shelf & turn them into something useful.

Process maps are the worst. I hate getting paid to make something that immediately goes onto the shelf, never to be looked at again.

And you know what every organization’s shelf is full of? PROCESS MAPS.

People love to make these damn things, or hire consultants to make them, and then bask in the sensation of accomplishment. Of having “documented”—or even worse, “understood”—the current state. What a waste of everybody's attention!

Next time somebody gives you a process map and shrugs, saying, “well here it is, although it’s a little out of date,” here are three things you can quickly add to it.

One: Clearly identified wiggle room

Even in the most dialed-in, locked-down business process there is some wiggle room. Find it. Circle it. Shade it in on the diagram. If 90% of the process is utterly inflexible due to agreements, requirements, contracts, deals, budgets, SLAs, suppliers, etc., there are still some things that can be changed. Find that 10% and make sure everybody else can see it too.

Two: Controls and why they’re there

Controls exist to mitigate risks. Identify the controls built in to this process, and draw them into the process map if they’re not already there. For each control, write down how often it happens, whether it’s automated or manual, what risk it’s there to guard against.

Sometimes processes have too many controls built in. Sometimes, not enough. How are you going to tell unless you map them out?

Three: Process behavior stats

For each step in the process, get some baselines. It’s OK to guess. Build a wee little table with these elements: How often does this happen? How long does this step take? What’s the percentage of time work enters into this process step 100% complete and accurate (%C/A)?

Now you know where the constraints or bottlenecks are. And you know where things break down. Remove constraints as best you can, build in quality where it’s needed. (Easier to not make mistakes in the first place than to have to sort them out later.)

TL;DR:

Process maps gather dust since they don't explain much on their own. Improve them by...

Identifying wiggle room or degrees of freedom, so people can see opportunities.
Listing controls, so people can argue about risks and mitigation.
Measuring process behavior, so people can see where work breaks down.

Options & set-based design

An effective lean method for eliciting & testing out the best, weirdest ideas.

It’s hard to know when to keep your options open and when to commit. Hang onto too many options for too long and you’ll spread yourself thin. But you might regret committing too early—what if the grass really was greener on the other side?

This is hard to do as an individual. It’s nearly impossible within a group. That’s one reason why groups spend so much time examining and choosing between alternatives. But this eventually gets in the way of doing good, and being good. As the Liverpudlian philosopher said:

Life is what happens while you’re busy making other plans.

When a group is facing a big decision about what to do next, the game is to preserve and explore options, without letting the upkeep of options get distracting or expensive.

Introducing set-based design

Set-based design is a method that’s worked for me and my clients. It’s a key lean method that often goes unused for weird historical reasons.

Here’s how it works:

A group starts with a broad, unproven set of options. These might range from the status quo to the wildest alternatives.
Then they will experiment, quickly kicking out ideas that won’t work.
Feasibility is evaluated in the open using simple, shared criteria. No one should be surprised by the results, even if their own idea is eliminated.
Eventually, the set of options narrows to one everybody can live with.
Take that one option (and maybe one backup) into development.

Benefits of set-based design

I’ve used set-based design to get groups through decision-making and evaluation processes, with some surprising benefits:

The method has helped risk-averse teams explore long-shot ideas.
Similarly, it has helped rambunctious teams commit to a single, sometimes boring, idea.
Set-based design is very fair about letting all stakeholders have their say and seeing their ideas given a fair shake.
The conversations about quality, cost, and value that are important for creating evaluation criteria have outlasted the design process. The local answer to “what good means” for a particular design activity can point to a larger answer about “what good is” for a team.

Starting points

Book—for theory and history: read Lean Product and Process Development by Ward and Sobek. Make sure you get the second edition. It’ll give you everything you need to get started.

Just want the overview? The authors published a decent intro back in 1999.

Paper—for trying it out: This 2016 CIRP paper lays out the exact approach that I’ve successfully used to plan set-based design activities for my clients.

April 16, 2024: edited for length.

Posts tagged with Continuous improvement

The 10x pancake-maker

The best pancake

Started but not finished

This is hard

The bad news

The good news

TL;DR:

First: See one

Second: Do one

Third: Teach one

You’re done!

For knowledge work

TL;DR:

One: Clearly identified wiggle room

Two: Controls and why they’re there

Three: Process behavior stats

TL;DR:

Introducing set-based design

Benefits of set-based design

Starting points