Make it easy to do the right thing.

That’s roughly how it looks from a management team, and I’ve sat in a fair number of them over the years. In most of them there have been a couple of items that never got done, that always came in late or came in at too low a quality, and that stayed on the agenda meeting after meeting without anyone really being able to explain why.

What I’ve come to is that what you see at that end is almost always a receipt for something else entirely, namely that further back in the chain it was easier to do the wrong thing than the right one. This text is about how you find that place, and about what I’ve seen work once you have. Part two shows how the route is built, and part three where it belongs in the bigger picture.

The door that squeaked

Nearly thirty years ago somebody told me a story and called it Toyota’s seven steps. I’ve used it ever since, and it was only now, sitting down to write it out, that I found out what it’s actually called.

A car brand had an unusual number of cars coming in for warranty repairs in January. Why? The doors squeaked. Why did they squeak? The lubricant had dried out. Why had it dried out? It was of lower quality than the old one. Why was the quality lower? They had switched to a different type six months earlier. Why had they switched? To save money. Why did money need saving? Head office had issued a savings target. Why had they done that? Because the owners wanted to improve profitability.

Seven questions, and the interesting thing isn’t that the chain is long. It’s that nobody along the whole of it did anything wrong.

The owners who want better profitability are right. Head office turning that into a savings target is doing precisely its job. The buyer who finds a cheaper lubricant is doing exactly what they were asked to do, and doing it well. Yet seven links later the cars are sitting at the workshop, and the person who has to take the call from the customer wasn’t part of any of those decisions.

The method is usually called the five whys, and it’s most often attributed to Taiichi Ohno. The name I’d been given wasn’t quite right, then. It actually comes from Sakichi Toyoda, Toyota’s founder, and it was Ohno who turned it into a method and coined the phrase about asking why five times. His own published chain is about a machine that stopped rather than a car door, so the story above should be seen as the illustration it is. And my ending up at seven instead of five turns out not to be my own invention, because there’s a well-argued case that Ohno compressed links at both ends and that the chain is really seven.

The method has drawn fair criticism too, among others from Teruyuki Minoura, himself a former managing director at Toyota, for being blunt and for making it easy to stop at the first technical symptom. The most common objection to an answer like mine is the opposite one, that seven links is one too many and that you’ve ended up somewhere nobody controls. I think that’s exactly where you should end up. That the chain finishes at a perfectly reasonable decision is the whole point.

One mistake, many people paying

I sold IT solutions for many years, and the same pattern exists without a car factory, of course.

A salesperson closes a deal. Good deal, right customer, nothing odd about it. But it never gets registered properly, or it gets registered without the customer’s purchase order number and without correct delivery information, which is to say without what the warehouse actually needs in order to send the right thing to the right place. The salesperson saved a few minutes and moved on to the next meeting, which was precisely what we’d asked them to do.

Then what usually happens happens. The warehouse guesses and ships, the goods come back and have to be shipped again. Finance issues an invoice that can’t be matched against the customer’s order, the customer can’t pay it, and the money comes in long after the due date. Counted in time and money it was never four minutes, it was several hours across three other departments plus one more shipment.

Then take an administrator entering customer orders. Some order types take two steps, others take five, and on screen the outcome looks the same because the order goes out either way. In the five-step version there are three steps nobody else sees, namely the discount that has to be applied, the free shipping that has to be flagged, and the amendment that has to be entered so the order carries the right information onwards. Skip them and the order still goes through. The discount simply fell away, the free shipping disappeared, and the customer gets an invoice that doesn’t match what was promised. Customer support takes the call, finance issues the credit note, and the salesperson has to explain themselves.

What the three stories have in common is that the person who caused the error never paid for it. The cost carries a multiplier, it lands on other people, and it almost never shows up in the same report as the saving. Correcting a mistake after the fact also costs nearly always more than spending a little more time up front would have, and the gap grows the further down the chain the mistake gets to travel.

Right doesn’t mean perfect, though, and that’s worth saying plainly. Right means right enough that the next person in the chain can do their job without first correcting yours. Where nobody further down is affected and where a mistake is cheap to fix, it’s often entirely reasonable to leave it. Chasing perfection in that sort of thing costs more than it gives, and it’s just as common a mistake as the opposite one.

Most people do want to get it right

So why don’t they get it right? Ask the salesperson and you’ll usually get one of two answers.

Either that it takes too long to fill everything in, and do you really want me making fewer deals and doing more admin? Or that it went so fast, I closed the deal in the car and then forgot about it. I’ve probably heard that second one at every single workplace I’ve been at.

It’s easy to dismiss those as excuses, and sometimes they probably are. But the vast majority of people do want to get it right and are simply missing the conditions for it. The salesperson has a target for customer visits and closed deals, and that’s what they’re measured and judged on. The best salesperson is also rarely the best administrator, and I don’t think I’ve ever heard anyone say they enjoy updating the CRM or writing quotes. They want to talk to customers and close deals, full stop.

Start from that and some of what we do looks rather strange. We ask them to spend time updating systems, building bespoke quotes and keeping price lists current, and then we decide that the quote can only be written at the office and that the CRM can only be updated from a computer. So we took something they already find tedious, and made it harder to get at.

Forty minutes to set up a customer

Sometimes it’s management that built the obstacle, and that’s the most uncomfortable part of this.

I’ve seen systems where it takes between twenty and forty minutes to set up a new customer, and roughly as long to create a new lead or opportunity. Why does it look like that? Because we wanted to be able to measure and follow up on a lot of things. Every single field was put there by somebody with a good reason, and together they became a threshold that no individual person ever decided on.

The awkward part comes later. After a couple of years working in the system you notice that a large share of what gets collected is never used. When we built it we thought we’d use it, then it never happened, and then nobody has had the energy to go back and clear it out. So the requirement lives on through sheer inertia.

There’s a name for exactly that. Taiichi Ohno described seven kinds of waste, and he put overproduction at the root because it feeds the other six. Collecting data that nobody consumes is overproduction in its purest form, and it generates precisely what you’d expect. Waiting, when somebody has to ask a colleague what a field means. Over-processing, when five steps do the job that two would handle. And eventually defects, when somebody skips the whole thing and the order goes out without what it needs.

The question to ask, then, isn’t how we get people to fill in the fields. It’s which fields are actually used, and what they cost across the twenty people filling them in every week. Four minutes a week for twenty people comes to around seventy hours a year, and that bill shows up nowhere, because it’s paid in small pieces by many people who each don’t think their share is big enough to raise. If the requirement doesn’t carry its own cost it should go entirely rather than be simplified, and that’s the cheapest improvement you can make, because the whole effort is a decision.

As long as you get away with it

Then there’s the other half, and it’s less pleasant to write.

As long as nothing happens when the steps get skipped there’s no real reason to spend time learning them properly. That isn’t unwillingness, it’s a perfectly reasonable conclusion drawn from what you see around you. And if the person who gets it right and the person who gets it wrong are treated the same, we’ve told them ourselves which of the two behaviours we actually mean.

But the order between these two things is decisive, and I think that’s where most people go wrong. It’s hard to ask others to take responsibility for obstacles we put there ourselves. First you remove the obstacles, because the vast majority genuinely do want to get it right. Only then, once the excuse is gone, is it reasonable to let the consequence land on whoever still takes the shortcut. Do it the other way round and it becomes finger-wagging, and finger-wagging is reminders in practice.

The reminder is the most expensive instrument a management team has, incidentally. A behaviour that depends on somebody having the energy to keep it up lasts exactly as long as that person’s energy does, and it costs attention every week, indefinitely. It also does something unfortunate to the relationship, because the person reminding becomes an inspector and the person being reminded becomes an opponent. Moving a field in a template costs an afternoon and holds for years, and the difference in price between the two isn’t marginal.

What actually bites is the consequence landing where the decision is made. That the salesperson who didn’t register the deal takes the call themselves from the customer who can’t pay. That whoever skipped three steps issues the credit note themselves. Not as punishment, but because that’s the first time you see the whole chain, and somebody who has seen it once rarely needs reminding again. The same holds in the other direction, exactly. What gets measured and what gets rewarded is what will get done, and if we only measure closed deals we’ll get closed deals, with or without what the rest of the company needs in order to deliver them.

The route doesn’t appear, it gets built

What I’ve seen work is undramatic, and it runs roughly in this order.

Start by removing the steps that don’t need to exist. Then automate what can be automated, so a piece of information is entered once and follows along by itself. Adapt the tools to how the work is actually done, so the salesperson can update the CRM from their phone between two meetings rather than at the office in the evening, and so the quote can be sent from the same place. Write down how it should be done, ideally as a checklist you can tick off, and put somebody alongside anyone new until it’s second nature. And make sure whoever leads does exactly what they’re asking of others, because nothing eats a decision faster than a manager who takes the shortcut themselves.

Best of all, of course, is when getting it wrong isn’t possible. A company car policy that sets out which makes, which value and which supplier you can choose between makes the decision easy for everyone, and nobody has to negotiate anything. Laptops and phones are ordered through a portal where only the right choice is available, and then getting it wrong becomes impossible without anyone having to say a word. That sort of solution is the most expensive to build and the cheapest to own, and it holds long after whoever built it has left.

The test

There are two questions I tend to ask to work out whether something is finished.

The first is whether the right route requires somebody to remember it. If the answer is yes, the behaviour rests on memory and willingness, and both disappear first when things get stressful, which is precisely when it’s needed most.

The second is what happens to the person who gets it wrong anyway. If the answer is that somebody else clears it up, quietly, a department away, then it isn’t finished no matter how good the instruction is.

Something to try

Take a symptom you actually see, meaning the returns, the credit notes, the late payments or that one measure that has slipped. Ask why, and don’t settle for the first answer. Keep going until you arrive at a decision that was perfectly reasonable when it was made, because that’s usually where the chain ends.

Then sit down with whoever actually does the work, in the real system rather than at a whiteboard, and go through the task step by step. Count clicks, logins, files that have to be hunted down and people who have to be asked, and at every step ask whether there’s any reason at all for it to still be there. It usually turns out that two or three steps can be struck out on the spot, and that one of them is the entire explanation for why the decision never held.

Finally do one thing that tends to be more uncomfortable than the other two. Follow a single case all the way through the company, from the first click to the invoice going out, and write down every person who spent time correcting something along the way. Put that total next to the four minutes somebody saved at the start, and show it to both of them.

In practice

In Genomlysningen, our review of the business, we look for exactly those places, meaning where the right thing costs an extra step and where the consequence of skipping it lands on somebody other than the person who skipped it.

The order that tends to work is to remove steps first, then move what remains to where the work is already being done, then make sure the consequence lands where the decision is made, and only after that to talk about who should remind whom. By the time you get there, that question has usually disappeared on its own. And where you can close the door entirely, so that getting it wrong simply isn’t possible, that’s obviously where you want to end up.

The next part is about how the route gets built, meaning the three stages of strategy, process and structure, and about why they have to come in that order.

If you’ve got a symptom nobody can quite explain, do get in touch, that’s exactly the kind of conversation I look forward to.