What does the customer want?

                    David and Sarah Kerridge

Summary

To learn how to improve service, we must know what will benefit 
the customer. The obvious way to find out is to ask: conduct a 
survey. Properly designed surveys are an important tool of Market 
Research, but surveys do not always give reliable results, and 
some even do harm. We give some of the reasons for this, and 
describe other sources of information.

Introduction

At a one-day seminar in 1989, Dr Deming was answering questions 
from the audience. One was:

     "How should we ask the customer what he wants?"

His reply was unexpected and thought-provoking.

     "Customer? What does he know about it?"

     This sounds surprising: after all, Dr Deming teaches us to 
regard as the customer as the most important part of the system. 
But he went on to give many examples which show that the customer 
rarely knows what he wants.

     For example, before developing the first photo-copying 
machines, the Xerox Corporation paid for a large, expensive, 
survey to estimate the market for office copiers. This market 
research said there would be no demand for them. As Dr Deming 
says, everyone thought Firestone was mad to develop pneumatic 
tyres, when solid tyres were obviously more reliable. There are 
many similar examples: if you wait until customers know what they 
want, you lose the customers.

The customer does not always know

In the Xerox example the customers were asked about something 
they had never seen. The demand did not exist until the service 
was provided. Is it any better when you ask about familiar 
things? Sometimes it is: if you ask strictly factual questions, 
such as which items have been purchased, or which services have 
been used, the replies may be less inaccurate. Even so, memory 
plays strange tricks. Customers systematically underestimate the 
number of cigarettes they smoke, for example.

     Market research firms have conducted experiments in which 
consumers put the packaging from every item used into a special 
bin, so that the total usage is accurately known. The results 
differ greatly from answers based on memory. Observed behaviour 
is always more reliable than stated opinion.

The sample must not be misleading

Quite apart from errors in reporting, it is important to get a 
good sample. If you rely only on customers who take the trouble 
to fill in and return forms, the results will be biased. The ones 
who answer may be the most helpful, or the least busy, or even 
the ones who like complaining. What is worse, you learn nothing 
about the most important group, the customers you could have had, 
but failed to get.

The questions must be carefully chosen

I never cease to be amazed at some of the questions asked in 
surveys of this type. For example, after a rail journey on a 
train that ran late and was far from clean, I was asked to fill 
in a form containing questions like:

     "Did the train arrive on time?"
     "Was the train clean?"

Do they not know what their trains are doing? Or are they trying 
to find out whether we notice when things go wrong? Perhaps both. 
It gives a very poor idea of their quality systems.

     When you ask for opinions, like "Would you prefer....?" or 
"Are you satisfied with...?" expect problems. Both experience and 
theory tell us that the answers rarely mean very much. To take an 
extreme case, suppose that you want to know what your own staff, 
as internal customers, feel about their conditions of work. You 
could design a survey, with questions such as:

Are you paid   A) too little  B) too much    C) enough

Do you work    A) not enough  B) too hard    C) just right

     The answers are so predictable that they tell you nothing. 
You might think no one would ask such silly questions, but I have 
seen some nearly as bad. One questionnaire asked, in a slightly 
concealed form, if customers would prefer lower prices. No doubt 
the aim was to find out whether customers would prefer lower 
prices, or better service. But why ask? Continual improvement 
will produce both.

Surveys may do harm
 
The concealed purpose of many customer surveys is to check up on 
staff. The size of bonus depends on the scores obtained. If your 
employees know that you are checking up on them, and asking 
customers to report on them to management, this will destroy the 
atmosphere of trust and pride of workmanship that you must strive 
to create.

     Another gross misuse of a survey is as a public relations 
exercise. A company once asked me to design and supervise a 
survey. When I became suspicious, and questioned with great 
persistence, the company admitted that the sole reason for doing 
the survey was to present a caring image. They had no intention 
of using the results. In another case, a hotel admitted that the 
"customer satisfaction" forms were largely ignored: but they 
allowed customers to express irritation harmlessly.

     In both these cases there is an attempt to fool the 
customers. Even if it succeeds in its aim, and we ignore the 
ethics involved, it does great harm to the organisation. It 
encourages an attitude to customers which makes sure that quality 
will not improve.

     This does not mean that customer surveys will never 
do any good. If there is something that irritates a large number 
of customers, you will find out fairly quickly. On the other 
hand, a really good survey is expensive and can not be hurried. 
This means it is difficult to use surveys repeatedly, as part of 
the PDSA cycle. It makes sense to see whether there are any other 
ways to get the same information.

Why do you want to know?

Before we start to gather information, we should spend time in 
planning. The first question is not what  do we what to know, but 
why  do we want to know. Ask first "what will we do with the 
information when we have it?".

     For example, managers sometimes say they want to know how 
well the organisation is pleasing its customers. Even if such a 
question can be given operational meaning, what difference does 
it make? The action, to work for continual improvement, should be 
the same whatever the answer. We should concentrate on improving 
all the processes. Provided these processes contribute to the aim 
of the system, the overall outcome will take care of itself.

     A good reason for wanting information is to identify trends, 
such as changes in customer behaviour, that affect planning 
for the future. For example, an airline must keep an eye on the 
balance between business and holiday traffic, or scheduled as 
against charter flights.

     Analysis of customer complaints by type can be useful in 
some ways, but total numbers can be misleading. A rise in the 
number often simply means that complaints are being dealt with 
more sympathetically. Something similar affects crime figures. 
The more effective the police force, the more crimes are 
reported.

     Information about potential customers sometimes must come 
from surveys. Information about existing customers can often be 
gained more cheaply and conveniently from sales or other records. 
Records will only be useful if they are very well kept, but 
excellence of records should be taken for granted in a quality 
organisation. In fact, if record-keeping is improved enough to 
make surveys less necessary, there will be many other benefits.

Concentrate on process rather than outcome

Although we obviously want to have satisfied customers, measures 
of this outcome may not be the most helpful information for 
making improvements. We must remind ourselves of the true nature 
of quality. Of course we must satisfy the customer. This just 
means that there are no complaints, but a policy based on 
avoiding complaints is too negative. A satisfied customer may 
be just as satisfied somewhere else.

     We should continually think of ways to increase benefit to 
customers, not just satisfy obvious needs. Delighted customers, 
who get more  than expected, not only keep coming back, but bring 
friends along as well. The way to achieve this is the continual 
improvement of every process which contributes to the final 
result, not just the processes directly visible to the customer.

     When the attention is focused on processes, rather than 
overall outcome, it is much easier to decide what information to 
gather. The numbers of mistakes in invoices, the time a patient 
has to wait, or the rates of staff turnover and absenteeism, are 
valuable in answering specific questions, and not difficult to 
study.

Ideas for innovation in service

Successful innovation depends on information, but usually needs a 
depth of insight into customer thinking that is hard to get from 
a questionnaire. In depth studies of special groups - a standard 
market research technique - may be more helpful. But the best 
source of information for this purpose is usually your own staff.

     Before designing surveys, make sure that all your staff are 
looking for ways to make improvements. The providers of each 
service usually have direct contact with customers: far more than 
the managers. They should be the first to know how customers 
respond. Managers are more likely to learn by being customers 
themselves. Everyone uses services. Managers of an airline are 
bound to travel themselves: they should sometimes fly economy 
class, or with a competitor.

     Completely new ideas are best tried out on a small scale, 
sometimes called test marketing. This one obvious advantage that 
service industries have over manufacturing: it is much more 
expensive to test market a new car. The design and analysis of 
such trials must be done on sound statistical principles.

Break down barriers

In most organisations, staff at all levels know of ways in which 
goods or services could be improved. Usually, unless the Deming 
philosophy has taken a firm hold, they do nothing about it. There 
are various reasons for this, all of which spring from neglect of 
Dr Deming's 14 points. Here are just a few:

     1)   Lack of communication between different departments, or 
          different levels.

     2)   The feeling that managers will take no notice - 
          "workers are paid to do what they are told, not paid to 
          think".

     3)   The fear that managers will treat all suggestions as 
          criticism.

     4)   The fear that improvements in efficiency will lead to 
          more work, without increased reward, or alternatively 
          less work, leading to redundancies.

          .....and so on. All of these are based on experience. 
Until you know what to look for, you are unlikely to see them. Or 
perhaps, until you know that such problems can be removed, you 
accept them as inevitable.

     Such barriers always exist, until transformation is far 
advanced. The process of finding and destroying them is a potent 
means of improvement. Just saying that suggestions from the work-
force are welcome is not enough. The members of the organisation 
must know from experience that any suggestion made will be taken 
seriously by the management, and either acted on promptly, or the 
reasons why not fully explained.

Conclusion

There is no doubt that surveys can be useful: in many cases 
essential. But never commission a survey without careful thought. 
Before you start, ask several questions:

1)   Could we get the information we need from existing records?

2)   If the information recorded is not accurate enough, can we 
     improve the records?

3)   Do our own staff already know the answers?

4)   If so, have we ignored what they tell us, are they afraid to 
     tell us, or have we not asked them? Do we not trust them?

5)   If they ought to know the answers, but do not, why is this?

6)   Would a survey give reliable enough answers?

     If, after all this, you are still sure that a survey is the 
best way to get what you need, make sure that it is designed and 
executed on sound statistical principles. Good surveys are costly 
and difficult, and expert help will be needed. In the right 
place, and correctly carried out, the survey is an important tool 
of continual improvement. A bad survey is worse than none at all. 
It will give false information, and lead to expensive mistakes.

     Make sure that you consider the psychological impact of 
everything you do. This may be greater than the obvious effect. 
While bad surveys can do harm, getting the whole work-force to 
think of improvements is bound to do good. It produces a state of 
mind in which service to the customer is seen to be important.