CUSTOMER SERVICE - Developing Successful Customer Satisfaction Survey

Introduction

In today's competitive business environment, quality service is often the difference between keeping customers and losing them. Good marketing brings new customers in, but great service is what keeps them. But to provide good service, you need a way to know how well you're doing - to find out what your customers think is important, what they want, and where you need to improve. And the best way to get this information is with a survey.

But you won't get useful information if you don't ask the right questions. What exactly do you want to find out from this survey? What areas of service do you want to ask questions about? You don't want customers doing the survey and saying 'I wish I could comment on this - why can't I?'

A good start is to conduct a few pre-survey interviews with customers, just to make sure that your questionnaire doesn't overlook any important areas. Also talk with your frontline employees, the ones who deal with the customers on a daily basis. Ask them and their immediate supervisors about the concerns their customers most often raise. This will give you an idea about what areas your customers are most concerned about, and therefore what you should ask in your survey.

Types Of Questions
Essentially, you can divide your questions into three types - and therefore, three sections.

1. Overall company ratings.
2. Rating of specific aspects of service, specific areas.
3. Demographics about the respondents.

You want to know what the customers think of your company in general- and then you want to ask them more specific questions about particular areas that you (or they) are concerned about. Survey software like Key Survey, with its branching and piping features, allows you to find out - while asking the general questions - what specific questions you want a particular respondent to be asked later. For example, you might be more interested in asking them follow-up questions about an area they rated 'very good' or 'very bad' than about an area they rated 'neutral'.

The third section, demographics, is necessary because you want to know which customers give which answers. Knowing who your customers are is helpful in itself - but if you combine that information with other answers, you can get a very specific and useful set of data about how different types of customers view different aspects of your business.

How To Ask Questions

It's been proven repeatedly that the way a question is asked can have an influence upon the answer - this is how political campaigns can generate opinion polls saying the exact opposite of what their opponents' poll said, for instance. For your business, however, you want a survey that gets the most accurate and honest answers - you want accurate and unbiased information.

Remember that the purpose of a survey is to gather that accurate information, and not to sell or further market to your customers. Despite the apparent marketing advantage this might have, this will not only bias the data you're getting (ruining the ultimate purpose of the survey), but it'll cause a certain percentage of your respondents to become cynical ('do they really care about the information I'm giving them, or is it just an excuse to sell me on more stuff?') and not complete the survey.

It's much wiser to make the survey's questions as neutral as possible, gather a larger amount of better information, and then to use that information to improve your marketing efforts. Aside from the fact that just being asked for their opinion will raise your customers' opinion of you, a survey shouldn't in itself be considered a marketing tool.

This means that questions should be asked in a plain, matter-of-fact way, not the usual marketing/advertising tone. For example:

NOT: "John Doe Company makes every attempt to please our customers in regard to technical support. How would you rate our success in these efforts?"

BUT: "How would you rate our technical support?"

The way you set up your rating scale is also relevant - specifically, because giving an odd number of response options allows for a center, 'neutral' response. An even number of responses, however - without that comfortable middle ground - forces a respondent to choose one or another of the middle options; does he slightly like or slightly dislike the service? 

As a general rule, it's better to allow neutral answers in the early questions (i.e. by using an odd number of possible responses) and then, on anything that follows up on those, to disallow those neutral answers (by using an even number of response options.)

Overall Ratings

The first part of your survey will usually have an overall question for each major customer service aspect, and one or two general open-ended questions (text answers, as opposed to multiple choice) at the end. The purpose of this is to get an overview of the respondent's satisfaction, and a broad picture of how your customer support is seen. And with a surveying tool like Key Survey, capable of branching answers, this can also be a first step to getting more information.

In an ideal world, your customers would have the time to give you all the information you want. In the real world, you don't want to take any more of their time than you need to. Therefore, you want to focus your questions to get the information of most value to you. This is where branching comes in handy.

With branching, you can set ratings of 'extremely satisfied' and 'extremely dissatisfied', for a given area, to lead to further questions (about why the respondent feels this way); if someone says that they're somewhat satisfied/dissatisfied or neutral, you can note that and not ask any further questions. This way, you can focus on what's really important to your customers - you can steer the questions asked to a particular respondent, to be the ones of most relevance to that respondent.

Usually the scale here is 'Excellent, Good, Average, Fair, Poor', or 'Very Satisfied, Satisfied, Neutral, Dissatisfied, Very Dissatisfied.'

Some typical questions of the level you might want to ask here might be:

'Please rate John Doe Company's overall customer service',
'Please rate John Doe Company's overall documentation',
'Please rate John Doe Company's customer support',

And so forth - general questions. A good rule of thumb is, one overall question per department or major service-area description.

This is also the place for asking customers to rate the importance of various aspects of customer service, and to ask reference-type questions, such as 'How likely would you be to recommend John Doe Company to a friend or a colleague?', and 'How likely would you be to use or buy [Product Name] again?'

Essentially, the first part of the survey is where you want to get an overview of how the customer sees your company. You can then use this to branch down into more detailed specifics in the second part.

Specific Ratings

Specific-area questions should comprise most of the survey - gathering detailed information about subsections of your customers' experience. For example, the single question related to 'tech support' in the first section, can be expanded upon here into multiple questions: what did the customer think about the wait time, the courtesy and politeness, the ease in which the problem was solved, the thoroughness, the accuracy, and so forth.

These ratings can then be combined to produce an average rating. Low scores can - using Key Survey's triggering feature - cause emails to be immediately sent to relevant people: "This customer is unhappy about your area. Call them."

They can also be used to rate specific employees - by encoding the customer information in the survey, for example, you can trace the results of certain questions to individuals. For example, you might find that of the customers who dealt with a member of Team Leader A's group were substantially more satisfied than those who dealt with Team Leader B's people. Some companies use this not just as a basis for where to assign training resources, but as a key part of employee performance evaluations.

Open-ended questions should also be used here - at least one per segment, saying words to the effect of 'Do you have any other comments regarding [area in question]?'

The information gathered in this section can give you a detailed and specific overview of exactly how your customers view your company - of exactly where your perceived customerservice strengths and weaknesses are.

Competitive Ratings

Asking companies to rate your service on an absolute scale is one way to do this. But your company doesn't exist on purely an abstract absolute scale; the context against which your customers view your company, and the context that you really care about, is your competitors. How good are you as compared to them?

This can be a helpful question to ask, especially in extremely competitive environments. In some cases, there are only two or three companies in a field, so you might want to ask about them by name: "Based on your impressions of Y and Z companies, how would you rate X's telephone support: Significantly worse, somewhat worse, somewhat better, significantly better?"

In other cases, you can ask on more general lines: "Compared to other companies in the field, how would you rate X's telephone support?"

It's also possible, and sometimes useful - if you think enough of your customers know something of your competitors - to ask companies to specifically rate them in order: "Which company is best at this? Which company is second-best at this?" This sort of information can give a very useful view of your competitive landscape.

Branching

There are two reasons you might want to use branching in the second part of the survey. The first is to ask more detailed questions about areas the customer has flagged specific happiness or unhappiness: say, if they said they were 'very satisfied' or 'very dissatisfied' about a given general area, you might want to ask them for details about their experiences in this area, and to find out why they're satisfied or dissatisfied.

The second reason is to single out the areas the customer considers the most important - as determined by rating questions in the first part of the survey - and to ask about those. A respondent is going to be more willing to answer detailed questions about topics that he considers to be important, than about ones he doesn't care about.

Email Triggers

The major purpose of a customer satisfaction survey is to gather aggregate information, but a very nice side-benefit of doing this with Key Survey is that you can also flag specifics. If a customer says they're unhappy, it's reasonable to assume that they're thinking about going elsewhere - a situation where giving them special attention might stop them from doing so. And if a customer says that they're very happy with you, you might want to contact them and ask for a testimonial.

Key Survey lets you do this immediately, without waiting for a survey to be completed. Specific criteria (particular answers or ratings, to particular questions), or combinations of criteria (more than two 'very dissatisfied' answers), can directly trigger an email to one or more people - account managers, for example, to give the unhappy customer special attention and make them happy again. These triggers can contain more than just the customer name and the response; they can also include things like the text comments that the customer has made.

This is a very efficient way both to put out fires and to solicit testimonials; a customer who gives very positive responses might not trigger an email to the account managers or service departments, but might instead trigger one to the marketing department, saying that this would be a good person to contact for public praise.

Demographics

Demographic questions let you analyze your survey results by segments and sub-segments of your customer population. They also, in aggregate, give you a pretty good view of exactly what your customer body as a whole looks like.

They include questions about geography/location, position title, frequency of their product use, frequency of service use ('how often do you call our tech support?'), sex, age, income range and highest education level.

Which questions you should ask a respondent varies according to the type of customers you're asking - you ask different sets of questions to personal and corporate customers. (When you're asking questions of a corporate user, things like their income level and age are basically unimportant, while their job title and type of business are considerably more important.)

If your customers include both businesses and individuals, branching is once again very useful; it lets you use a question like 'Do you use XYZ personally, or as part of your job?' to decide which set of questions should be asked. Alternatively, you can use two different surveys: one intended for personal customers, one for business customers.

Survey Length

In theory, you'd like to ask all of your customers everything. In practice, this is impossible, because the survey would take two hours and nobody would take it.

It's possible to lie about how long a survey will take - 'this survey will take five minutes', when in fact it takes fifteen, in the hope that most respondents will, after five minutes, go through to the end anyway. Some market research companies (telephone and doorto- door interviewers) make a habit of doing this - anonymously, when often even the interviewers don't know who the client is. This sort of dishonest behavior is not something you want your customers to associate with your company.

Most good customer-satisfaction surveys take less than fifteen minutes to complete, and a significant majority of those are well under ten. Above that, you usually need some sort of incentive - although an incentive to do even a short survey is a good idea.

This makes branching a very valuable feature: you can focus in on the questions that are most important. You can cover a broad range of potential ground and general impressions in the first section, and then zero in for valuable specific information in the second.

The ability to autofill (where answers that you already know for a particular customer, are entered in without the customer even seeing the question) can be very useful, especially for demographic questions where you might already have that information on hand. (You can also use autofill to allow customers to check and correct demographic information that you have.)

Always tell your customers roughly how long the survey will take. Test it on a couple of co-workers who are unfamiliar with the questions. More respondents will be interested in taking the survey if they know how long it will take. A progress bar - showing just how far through the survey that a respondent is - will reduce the number of people who begin the survey but don't complete it.

Response Rates (And How To Increase Them)

It's nice to think that most of your customers will happily take your survey; after all, the information is to benefit them, right? Unfortunately, some customers don't have the time; others will earmark it for later and forget about it; others will start it and get called away.

10-15% is about the average response rate (surveys completed, out of surveys offered) that most companies get on customer-satisfaction surveys. There are a few things you can do to increase this, however.

One thing is a Key Survey feature called 'email reminders'. Essentially, if you're sending out a survey by email, Key Survey can keep track of who has responded to it and who hasn't. Those who haven't, will be sent another email with their survey link (or the survey enclosed, whichever you choose) and a message, after a set period - a week, say - to remind them that the survey is out there. This helps with the people who'd intended to do the survey but had forgotten about it.

Offering some kind of an incentive also helps: 'Complete this survey to get a coupon good for $X off your next purchase' or to enter a draw for something. A certain percentage of customers will fill out your survey just to get the incentive.

Sample Size

Any information is good. However, a survey that looks at aggregate information does need a certain amount of respondents - of data points to make an aggregate from - in order to be of real value in terms of analysis.

Statistically, you want at least thirty respondents in any subgroup you're analyzing. The more the better, of course, but below about thirty respondents, one or two particularly biased individuals can distort the results.

Analysis tools, by definition, let you find how many people answered a given question, and in a given way. If you're looking to analyze the responses of a particular subgroup (say, personal-use customers aged 25-39), and there are less than thirty respondents in that subgroup, you might want to be careful about assigning too much significance to the results. Alternatively, you can broaden the search criteria to draw your analysis from members of a wider subgroup - perhaps a neighboring one.

For example, if you only have 22 respondents in the age 25-39 subgroup, but you also have 17 respondents in the age 18-24 subgroup, you'll have a better sample - and more valid data - if you conducted an analysis on the results of people aged 18-39.

Conclusions

Surveying your customers is the best way you have of getting valuable information about who you are and how your company is doing. You'll get valuable information for improving your service, your product lines and your marketing; you'll get information that you can use strategically and for HR purposes.

Electronic surveying is the cheapest and most efficient way to do this, because you avoid expenses such as phone-interviewer wages and data-entry, as well as the delays caused by mailing. Higher-end electronic surveying options (not necessarily higher-cost ones) allow you a range of useful features such as branching, autofill and email reminders, which help you write more efficient surveys and increase the response rate.

Make your surveys short, detailed and to the point. Your respondents will appreciate it - and with the information you gather, so will your bottom line.