Seddon petition – politics worth voting for

The inimitable John Seddon is being proposed as a new UK ‘Public Services Tsar’.

His summarised manifesto:

1. Shut down the Audit Commission and other inspection bodies.

2. Get rid of the thousands of Whitehall target setters and rule makers. Retrain them to do something more useful instead.

3. Make public sector managers responsible by freeing them from compliance with Government edicts.

Greater detail here (.doc).

John Seddon

I can’t imagine a more powerfully positive (and – let’s face it – entertaining) outcome for British public services.

If you agree – and you’re a British citizen – then please do sign the petition to back his appointment.

If you haven’t come across this forthright thought leader before, then I highly recommend a read of his back catalogue (see the ‘Go Forth and Read’ panel to the right). This is a good place to start.

Misapplication of science to leadership

Here’s another reason I suspect may be behind why Western business didn’t quite ‘get’ the empowered workforce aspect of Japanese lean approaches:

It doesn’t feel very scientific.

Our logic follows ’cause and effect’ rules and some assumptions feel as immutable as the law of gravity.

• We think that documenting a procedure reduces the risk of it going wrong.
• We believe that introducing a bonus will improve performance.
• We know that separating work into its smallest parts will make it easier to learn and produce an economy of scale.

 

Bad Science (Ben Goldacre) by David Bailey's Son (cc)

The concerns a command and control manager would express about my suggestions for bringing about real empowerment would be:

• “If you let front line workers make decisions, there’ll be inconsistency and chaos.”
• “People will slack off if you give them freedom to decide how to work.”
• “Concepts like targets, SLAs and bonuses provide control.”

The problem here is that we’ve let long-held assumptions about work performance become unstated ‘law’.

What we should be doing instead is developing new hypotheses and testing them.

Those new hypotheses should come from mapping what’s happening in the work today.

But for that to happen, leaders need to completely change how they operate.

TPS: The leadership problem

Reading about the West’s adoption of lean principles – and specifically the Toyota Production System – leaves you with the distinct impression that there’s something missing in the leadership.

We got the techniques. We got the tools. We heard the words about empowerment and involvement, but we didn’t really get the meaning. Why?

I suspect it’s partly because heirarchical management is so deeply ingrained in our history and culture. Our entire system is based on the premise of authority.

Want something done? Announce a new rule. Laminate it. Frame it and put it on the walls. Reward people for following it. Penalise people for deviating from it.

In an attempt to engage people, we might try conferences, team huddles and suggestion boxes, but it’s all to little effect – because the thinking hasn’t changed.

 

Protest March by infomatique (cc)

In order to achieve real empowerment and to maximise what lean thinking can do, Western leaders must accept that:

• Front-line workers should make the tactical decisions about work, not managers.
• Those decisions should be based on representative data about work.
• Tacit knowledge plays a huge role in the work – documentation has severe limitations.
• Motivation is intrinsic to well-designed work, not extrinsic.
• The vast majority of people want to do a good job, even without any other incentive.
• Targets and Service Level Agreements make performance worse, not better.

Is CRM just covering over cracks?

Are you thinking of implementing a Customer Relationship Management (CRM) system? Look carefully at the cost/benefit case.

Such software gives an operation a way of logging and retrieving records of all activity with customers.

That’s very appealing to an organisation that has departments working in silos, fragmented processes and frequent failure demand. If a customer’s going to be passed from pillar to post, it’s better that they don’t have to repeat themselves with every new contact.

 

Handshake by oooh.oooh (cc)

But look at it another way: why not save the cost of the tech investment and the ongoing pain of your customers and instead re-design your processes and structure around the customer demand?

If you can deal with a customer in one transaction, they won’t care if you remember their birthday, or what they asked you for last month. You’ll save the cost of handling their repeated contacts with you. Your staff will get greater job satisfaction. And you won’t have spent a cent on additional technology.

It’s an angle you won’t find in the sales brochure, so just make sure you get an objective cost/benefit assessment.

When service measures lie

Do your speed of service measures tell the whole story?

Just last week in an office across town, a very modern fairytale unfolded…

 

Hot Potato by cmatera (cc)

Day Zero: A new case is created in the front office. The front office don’t handle these cases so they assign it to the back office.

Days One to Four: The case sits in the back office queue as they’re working to a backlog.

Day Five: The back office staff realise the case needs technical support, so they transfer it again.

Days Six to Nine: The case sits in the Technical team’s queues, as they’re working with a backlog too.

Day Ten: The Technical team work on the case, then send it to the Board for sign-off.

Days Eleven to Thirty-One: The case sits in the Board’s in-tray because senior management have to fit in a lot of terribly important meetings (and occasional rounds of golf), plus nobody questions the age of their work queues.

In the meantime, the customer who started the whole chain of events has called in three times to chase progress, adding a further 3 ‘cases’ to the system.

All in all, the case has taken a month to complete, but nobody knows that because each individual operational unit has met their own internal targets for getting cases off their own desks.

You might think this is an exaggerated example, but I’ve come across far worse. If you strongly enforce a turnaround target within units, teams or even for individuals, the problem is magnified.

To avoid falling into this trap:

• Only measure from end-to-end – that is, when the customer would deem the case begun and closed.
• Put your measures into a control chart so you understand the predictable variation.
• Work on the causes of variation until your process is in control.
• Remove the time-based targets and don’t inspect old cases. Focus everybody on the previous 2 steps instead.

How military tradition led to Dilbert

The ‘command and control’ working methods of many companies today have their roots in ancient military history.

While that was appropriate in the past, as industry enters a phase in which knowledge and creativity count for more than manual labour, it’s not going to work for much longer. 

So if your office life ever resembles a Dilbert cartoon, it could be that you’re feeling the effects of this approach on modern work. Here’s what I’m talking about..

Armies use public awards as a form of extrinsic motivation:

Pay is contractual – footsoldiers don’t share the spoils that their commanders gain from the army’s effort:

Decision making is restricted to the leaders, assuming that their decisions will be superior to those of the lesser-educated, lower-class ranks:

Working life is run according to strict rules:

The only route to greater benefits and social status is to climb the hierarchy:

The CEO of Cisco says that companies sticking with the command and control approach have only 5-10 years left before extinction. There’s a better way to do things.

Stop with the cheer-leading

You there. The service manager. Yeah, you.

Ever woken up with a cold sweat in the middle of the night with the sneaking suspicion that maybe all of those motivational talks that make you feel ever so slightly like David Brent might be… a waste of time?

Well, it’s time to apply a little scientific thinking and find out.

Mountains of cash has been spent and plundered preaching the art of ‘coaching for performance’. Consulting firms have sold glossy variations on the same theme since the late 1980s:

“Think your way to better performance;” ”Success is an attitude;” or – and it pains me to type it, the utterly ridiculous Law of Attraction.

Some service companies dipped a toe into this thinking – maybe with an annual conference to gee up the troops or some spirited words at a performance review.

Others snuggled the thinking to their corporate bosoms, using twice-daily huddles to maximise performance by asking staff “What will you do to be AMAZING today?” and later “So how did that pan out? And what will you do to be EVEN MORE AMAZING tomorrow?”

Now, you may have inferred from my tone that I think this is about as sensible as trousers made from herring – and you’d be right. But let’s bear in mind the fact that I’m a cold-hearted cynic and that you should never trust anything you read in a blog.

Instead, to make sure that your breath isn’t being wasted you should measure the impact of your actions.

Let’s say you’re hoping that pep talks will increase the number of sales leads your team will produce. Rather than doing the motivational talks and then cheering/booing the team based on the result, try plotting the number of sales leads they achieve over time.

Don’t average anything, just create a scatter graph and let Excel calculate the mean and Standard Deviation for you. Obviously, you’ll have a lot more data than I’m showing in this illustrative control chart:

Sales Leads from our fictional team - a Control Chart

Those Control Limits represent the mean plus or minus 2 Standard Deviations.

In short, this chart is telling you that in the current system, there’s a 95% chance that nothing the team is doing will produce more than 44 leads (or less than 17).

So maybe without the team talks these limits would fall. So test it. Go for a few days (or ideally weeks) without the wooping. Don’t change anything else. Re-plot your data.

I’ll pay you 10 of your American dollars if performance is statistically different.*

 

*Terms and conditions – you must be able to prove that the absence of coaching talks was the only difference between the 2 time periods and I’ll need testimony from a quarter of your team to prove they weren’t in on it. Do all that and I may just eat my hat, too.

Anatomy of a Control Chart

If you’re managing your service operation through tables and bar charts, you’re missing something.

Replace them with a control chart and you’ll see another dimension of your operational system’s performance.

Let’s say you process loan applications and you aim to get from first application to transferring the funds within 7 days.

First, ditch the target. It isn’t helping you.

Next, sample at least 1% of all completed cases and plot the data in a control chart:

 

Loan application control chart

In this example, I’ve used the average turnaround time sampled per day in order to see trends over time, though you should also plot the raw data to eliminate the averaging effect.

Interpreting a control chart

1. The average.

The mean of all the selected data points gives you a handy yardstick of how the processes is coping. But it’s not as interesting as…

2. The variation.

The upper and lower control limits show you the mean plus and minus 2 standard deviations of the mean. That’s a statistical way of showing you the extremes of your process performance within a 95% confidence level.

The loan application unit can be pretty certain that the next case they process will be set up at some time between17.5 and 41.8 days.

3. Outliers.

Outliers are data points that fall well outside of the control limits.

It’s something of a shame it draws the eye and that in many businesses, everybody spends time worrying about what caused them. It’s interesting to know, but essentially their existence means that something other than the norm was taking place to produce them.

As such, you can get much greater returns on your time by understanding what causes the performance between the control limits. Get to the bottom of that and you can make improvements that’ll help every single case in future.

 

Monitor trends over time

4. Trends and changes

If you’re successful at identifying the root causes of variation and delay, your control chart may start to look a little like the one above. For each clear step change in the performance, you should re-calculate the mean and control limits.

As a general principle, you should aim first to reduce variation, even if the mean stays the same. Once you have a more predictable performance, you can move on to trying to move the mean in the right direction.

This kind of visual representation lets you see the difference between real, lasting process change and predictable variation.

Are you convinced yet? If so, can get started with Excel’s chart and calculation functions. If you’re going to use the charts extensively, you might be interested in the software that does all the hard work for you. Winchart works for me, but there’s also Signals from Noise and a whole host of others.

If you work with control charts, I’d love to hear your tips and advice in the comments.

Process variation – handle with care

Throughout the 20th century, manufacturers – and more problematically, service companies – have been working towards that state of nirvana where all processes are as simple and uniform as possible. Why? Because received wisdom has it that this is the cheapest way to do business.

That’s true – if your demand looks like this:

Consistent demand, simple processes. Hurray!

A request comes in for Widget A, and Widget A goes out.

Does your demand look like that? Because in our experience, demand tends to look a little more like this:

Complex and variable demand

The variation in demand could occur with requests for product feature tweaks, special circumstances around a standard request, or perhaps a need for a product or service that you don’t typically provide (or that doesn’t exist in the market – and therefore a commercial opportunity).

When you try to apply simple, standard processes to demand that varies like that, you get a result that might best be represented like this:

Complex demand fed into standard processes

Messy. But more importantly, expensive. All of that ‘mopping up’ that has to be done as a result of badly-placed simplification is costing you time, money and customer goodwill – you just don’t see it because your measures don’t demonstrate it.

That’s the good news. It’s good, because it means you have scope to start reducing costs in all kinds of places, just by thinking about the work in a different way.

Here’s a quick overview of how to do it:

• Get meaningful data on what’s happening predictably.

Measure the types of demand that are placed on the operation, not just activity. Get a large enough sample in order to have confidence in which are common and which aren’t.

• Be selective in standardisation – automate the truly consistent.

Only standardise where the data shows that the input and output of a given process are consistently predictable. That doesn’t mean there can’t be more than one input or ouput type, or that you can’t have multiple different standardised processes to meet a lot of different but predictable demands.

However, it helps you avoid the trap of applying  a ‘one size fits all’ approach to varied and complex demands. Further to that…

• Remember that machines don’t think.

Review your automated processes to see how many ‘exception’ cases fall out of otherwise standardised work. It could be higher than you think – in which case you could actually save the time it takes to handle these twice by having a human handle them the first time.

• Chaos doesn’t need to reign.

Removing a standardised process doesn’t mean that a blank canvas and wild creativity has to take its place.

Once your front line teams have good data on the demands that don’t fit well into standardised processes, with some guidance they can still take an organised approach to develop their own alternatives and share the best ways they’ve found to handle them.

Just remove the obstacles that handcuff them to the old standardized version, whether that’s process documentation, system constraints, needless authorisation steps and so on.

• Refine the design – and keep measuring.

Got an idea for improving your standard process? Set up a controlled experiment to prove scientifically whether it’s more efficient in practice. Only implement major changes across the operation once you’ve got some good data to show that it’s better for your customers and cheaper for you.

As you start to design more and more of your operation to accommodate predictable variation, you will inevitably see complaints, duplication and cost go down – while your team’s empowerment and motivation levels go up.

Need help? We can work with teams to help this process transformation happen, or train business leaders to do it for themselves. Contact us for a quick chat about your operational challenges.

When standard processes cost more

As promised, here are some examples of misapplication of the ‘standardised work’ concept and its unintended consequences.

“You want a WHAT?”

This week, I needed to get a list of bank transactions from a 2-week period in 2007. It took 3 attempts:

1. I happened to be in my branch, so I asked the customer service desk for a printout. They told me that it was new policy not to issue them, so as to encourage customers to bank online. I was already registered, so decided I may as well print them out myself.
2. Logging into my online account, it emerged that only 6 months’ worth of transactions are available for viewing, presumably to save on server space/data runs.
3. I called the service centre for help. I was told that I needed to wait 3 weeks for a duplicate bank statement and that a charge of a few Euros would be applied to my account for the inconvenience. I needed them urgently, so after a long conversation involving not an inconsiderable amount of begging, I resorted a threat to complain in writing.

At this point, the service rep admitted that she had the transactions I needed on screen and that, even though it was strictly against bank policy, she would print them out when management weren’t looking and post them to me that day.

How bizarre. What’s happening here?

My First Clones by natpie (cc). Standard isn't always good.

The bank’s primary motivation is to cut costs. They’re aware that transaction statement requests are a frequent demand and the prospect of eliminating the time and money spent on them is like a giant, flashing dollar sign.

They’ve standardized a procedure to try and reduce the demand. The thinking goes: we can dissuade customers from asking for statements by refusing, encouraging self-service, charging and deliberately having a long turnaround time for issuing them.

There are 2 faulty assumptions here.

1. They’re assuming that customers request statements for kicks.

“Mary – I know what we should do this weekend. We should get a nice Rioja, order some bank statements from 1996 and reminisce about that time we settled the electrical bill by cheque rather than Direct Debit.”

Customers don’t want to spend time contacting you when they could be playing with their kids, eating a sandwich or even grouting the bathroom. I can’t imagine that aversion techniques successfully deter many of them.

2. They’re not counting their costs properly.

Their calculation:

(Number of requests for statements) X (Time taken to produce and issue one) = (Cost of handling statement requests).

What they’re missing is:

* The time taken to handle repeated attempts by the customer to get what they need. In my case, it cost the bank resources in dealing with my branch request; in bandwidth when I logged onto my online account and in call centre rep time when I had that final, overly-long conversation.

* The time taken to handle complaints. Had the service rep that took my call not contravened company policy in order to do what she felt was the right thing, I would have escalated the matter to her manager and possibly also sent in a written complaint to the bank. Bank’s interpretations of their regulations (such as they are) require that complaints are handled in a reasonably bureaucratic way. The result would be entirely avoidable cost.

* The reputational damage and loss of goodwill. I think we can assume I’m not going to recommend this bank to my friends. Perhaps worse is the impact on the service rep who took my call. How will she feel about her employer, when they effectively require her to break the rules in order to help people?

Here are some other examples where standardisation can have hidden costs:

• Web forms that demand a UK postcode (or US zipcode), when global consumers don’t always have one.  Are you losing sales like this?
• First level IT tech support that can’t help if the solution to your problem is not ‘have you tried turning it off and on again?’. The moment that the call’s handed off to the next level, the first transaction became pure overhead.
• Badly-designed telephone IVR options. ‘Press one to buy our product. Press two to pay a bill. For all other enquiries, go to our website.” You might think you’re saving money by discouraging demand, but if a customer wants to contact you, they’ll find a way. And they will take a lot more of your time to tell you about it.

Perhaps you know all this, but your company can’t afford to reduce standardisation. Bespoke service costs more, right?

In the next post I’ll suggest some ideas to try and convince you that it doesn’t have to.