Measuring Customer Satisfaction

In my last post, I made the claim, “the principal product of all inbound call centers is the mass production of customer interactions… In inbound call centers, the object of the interaction is to satisfy the customer’s reason for calling.”

If this is the purpose of spending a great deal of money on equipment and personnel, there must be some way to measure how well the center is meeting its goal of satisfying customers.  I would like to look at some of the ways that companies measure this.  But first, why should a company worry about satisfying their customers who call?

In an article in the Harvard Business Review of September-October 1990, Frederick F. Reichheld, a director of Bain & Company, and W. Earl Sasser, a professor at the Harvard Business School, described the need for quality to be measured for service businesses.  The article is titled “Zero Defections: Quality Comes to Services.”

The main thrust of the article is that the longer a customer remains a loyal customer to your company, the more profitable it is to provide him/her with your service and products.  To figure out how to provide the kind of quality service that is appropriate for your customer base, you need to figure out why customers “defect”, which means that they have stopped being your customer and become one of your competition’s customers for the same or similar services and goods.  They have apparently measured the difference that reducing defections makes: “Companies can boost profits by almost 100% by retaining just 5% more of their customers.”1

In addition to some very impressive examples and anecdotes in the article, there are some important concepts that I try to bring to the attention of my call center clients.  One of these concepts is to supply superlative service to all customers for whom it is cost effective to do so.  “Achieving service quality doesn’t mean slavishly keeping all customers at any cost.  There are some customers the company should not try to serve.  If particular types of customers don’t stay and become profitable, companies should not invest in attracting them.”2

The leading concept, though, is really a process: learning how to identify when a customer “defects”, and then following up with asking him/her why they are leaving.  “Unlike conventional market research, feedback from defecting customers tends to be concrete and specific.  It doesn’t attempt to measure things like attitudes or satisfaction, which are changeable and subjective, and it doesn’t raise hypothetical questions which may be irrelevant to the respondents.”3

So, the message is, learn to provide better quality service to your customers, learn how to retain them by asking those who defect why they are defecting, and by delivering better quality service, you will be more profitable in the long run.  Surely a good reason for the call center to satisfy customers’ requests.

Measuring customer satisfaction is usually done in one of several ways, though based on the comment quoted above, the results are probably transitory and may not be reliable.  There are market research firms that will perform customer satisfaction surveys for companies for a fee, and these may be the more reliable way to obtain that information.  They have experience with the statistics that are used to compile the information, and they have experience asking questions that should elicit good data.

Another way that companies measure customer satisfaction is to ask callers or web users if they will take a survey at the end of their call or session.  There are some limitations or shortcomings of this method, which may apply to a survey conducted by an outside company as well.

The first limitation is that of the group that participates in the survey.  The entire pool of customers for a company will never be sampled: not everybody will call during the period for the survey, nor will everybody who is asked to take the survey be willing to do so.  One would like to think that those who agree to answer the survey questions would be a random sample of the pool of customers, but it may not be.  For instance, if the survey period started, then there was a product recall which affected one of fifty different products, a high percentage of callers would be those who need to figure out how to return the product.  Would they be biased toward dissatisfaction, because of the inconvenience, or grateful because they were being taken care of?  In fact, those who might respond to the survey request could be a large proportion of those with an ax to grind.  Would that represent the real level of satisfaction among the company’s customers?  Not hardly.  So, there are potential biases that could show up in the answers.

The next limitation is to insure that the sample size is significant.  If only three callers of the ten thousand last month took the survey, is that enough to make inferences from?  Not really.

A third limitation, which would be less likely if an outside service ran the survey, is what are the questions really asking?  In the last post, I mentioned my experience with surveys which asked me to rate, on a scale of one to five, with five the best, whether the agent that I had just spoken to was enthusiastic.  Huh? There have been yes/no types of questions, such as ‘did the agent you just spoke with answer your questions?’  Well, yes, but if the answers were ‘we can’t do anything to help you’, what has the survey measured?  My satisfaction: not at all.  The agent’s performance? Maybe.

The questions are invariably asked from the company’s point of view, and actually sound like “we are not really competent to judge the performance of our agents, so we need your help to manage them”.  Excuse me.  I charge for that.

What is important to a caller when they call a company’s call center?  They have a question in mind that they want to get answered; they have a problem operating the product they have purchased and need help; they have a repair request; their bill doesn’t add up since there are charges they don’t recognize (or remember); they purchased something, it is the wrong size or color, and they want to return it for one that is the right size or color.  And, since they have approached this with a serious desire for resolution of their reason for calling, they expect to be taken seriously and treated with respect.  They can even expect that if they have purchased something from the company, they should be treated with a measure of gratitude.  If their problem is resolved and they are treated as if they deserve respect, they will be satisfied.  Seems simple.  So how do you measure this?

Basically you cannot measure it directly.  As stated in the Reichheld-Sasser article, satisfaction is an emotional state that may be fleeting at best.  So instead, what should be measured is customer loyalty which is the opposite of customers defecting.  Evidence of loyalty is provided by further account activity by customers: when a customer places another order, requests service again, or for some reason or another makes another contact.  By tracking the time between contacts, the average time can be derived, and when customers do not re-contact within that period, follow up offers can be made to insure that they have not defected.  In a peculiar way, loyalty may be measured by paraphernalia: if you sell baseball caps, t-shirts, bumper stickers, etc., loyalty can be visible.  However, all this requires either inferring loyalty or patience or a bit of both.

What makes sense, it seems to me, is to measure those activities over which the company and its agents have some control.  An individual call center agent has control over how much time they are at their desk and available to handle calls; to a certain extent, they can control the length of calls; and they can perform the transactions correctly.  Those can be measured, as well as how many calls were handled within given timeframes.  Likewise a company has control, to a large extent, over the number of agents scheduled to be available, which will bear on how quickly calls are answered, and the business process available for performing callers’ transactions.  These can be captured as data which can be turned into the information needed to manage the call center and the business.

Additionally, a well managed company will also track defections: people who say they are going somewhere else can be asked why, people who refrain from contact can be contacted with offers, etc.  Knowing if and why people defect can provide guidance for change if necessary.

Most companies that record transactions will announce that in their recorded messages, and will include the phrase “for training purposes”.  If that is the only way that the recordings are used: so that an agent can hear their own words, the inflections and tones of voice to better their performance, that would be a reasonable use.  But when the recordings are scored and used in performance reviews, the potential to use the recordings abusively and punitively rises considerably.  Add to that the possibility that however carefully the evaluation is done, the resulting evaluation will be perceived as not being objective, and the use of recordings becomes a risky proposition: risky because of the possibility of abuse.

A couple of thoughts here: in a  earlier post, I used a quote from Theodore Porter describing objectivity within business as a substitute for trust.  If the morale of the agents in a call center has value, which I believe it does, the quickest way to kill morale is to have management appear to be favoring some people, whether worthy or not.  My experience in call centers shows that when agents feel that they are being treated equally, fairly and objectively, morale is better, and the over-all performance is better.  If evaluations of recordings can be done fairly, with a set of performance goals that the agents all know and understand, and the possibility of being able to appeal evaluations in place, evaluations of recording may not detract from agents’ morale.

Another thought: the measurement methodology of the physical sciences has been set up as the touchstone of how to measure scientifically.  In an attempt to prove something in a completely different realm of human endeavor, often an analogy to the method of measurement from the physical sciences is incompletely or inadequately adopted.  Examples of this were discussed in the post about Stephen Jay Gould’s book, The Mismeasure of Man, where, for instance, Gould describes the great pains taken to measure skull size and draw the inference that the differences perceived in skull size meant that there was a physical reason for Afro-Americans were not at the heads of corporations.  In that case, the appearance of being “scientifically measured” gave a kind of credence to the “conclusions”, which were really foregone conclusions with the data actually cooked to support them.

The reason that I mention this is that I worry that under the guise of “scientifically measuring” performance in call centers, something else is going on.

When one approaches measurement of the natural world scientifically, the intent is to learn how whatever is being measured fits in the general scheme of things.  The intent usually is to be able to use the results to build a model sufficiently accurate that the model can be used to predict the future behavior of whatever has been measured within its context, and in some cases, to be able to control its operation.

When applied to measuring human performance, if the performance is in the realm of sports, it is used to record results, from which to improve.  When applied to individual human performance in business, measurement is used to control the behavior of people to increase the likelihood of profitability, based on some model of behavior that will be successful at generating profitability.  However the model of behavior was developed, it is taken to be the standard to which all the agents are held.

The potential biases in the human measurement are so large and the control often so much the guiding principal, that there can be the appearance of trying to hijack people’s individuality, to produce robot-like reproducible performance on the parts of the agents.

To tie this back to other discussions, this is one side of the tension described by the two French theorists, Constant, the pessimist, who worried that uniformity would “… suppress difference of thought and custom.”4

Since I have seen evidence indicating that Constant was correct in his concern, I too am concerned.

1 Reichheld, Frederick, and Sasser, W. Earl, “Zero Defections: Quality Comes to Services”, Harvard Business Review, September-October 1990, p. 105.

2 Ibid. p. 109.

3 Ibid. p. 109.

4 Ken Alder, The Measure of All Things, Free Press, a division of Simon & Schuster, Inc., 2002 New York, NY p. 317

This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s