How well is your customer service operation doing?

(What do your customers really think?)

(c) Aleks de Gromoboy 1997


Looking for an easy way to measure how well your customer service operation is performing?

I've set up satisfaction surveys for a number of customer service operations. The bad news that there isn't one easy measurement you can make. If you do go for excessive simplicity it may hide as much as it reveals. However getting good value out of customer surveys isn't that difficult. You need to identify what you want to know and the key factors contributing to those goals. Then look at both the average satisfaction against each factor and the spread of responses. Some responses saying you're excellent and some poor is not the same as all users saying you're OK!

The graph below illustrates the kind of spread you often see. It shows results of a helpdesk survey where the average score was a bit below acceptable - but there were clearly two different types of users. By focusing on the lower segment the helpdesk substantially improved the average perception. If they hadn't recognised the two types of user they might never have acheived acceptable results. (The lower segment was the group of users who had home PCs so had some expertise already.)

Some key considerations:

1: Identify the goal of the survey

There are a number of common survey goals. Typically these include: Identifying weak spots, noting trends from the last survey, comparing with industry standards, testing the results of specific changes, finding out how satisfied customers are with you, proving that you're meeting specification. All these have different needs.

Most important is to beware of measuring aspects that you can't or won't be able to change. It can be a waste of money and time.

If it's customer satisfaction you're looking at do you know what's important to your customers? How do you know? One of the classic mistakes of customer surveys is measuring the wrong thing.

There are a number of ways of identifying customers' real needs. Don't just rely on gut feel. Qualitative information can be found through through a small preliminary survey or focus groups. You'll get a mix of factors - identify the ones they consider most important (say the top 5). There are a number of techniques for identifying their priorities: e.g. asking for a top 10, to allocating 100 points amongst the categories, pairing (if you had to choose between this and that e.g. speed of answering and technical expertise) which would you prefer, having a list and ask them to score each item.

For example - one place I surveyed were sure that they did a great job because calls were answered quickly and problems were fixed quickly. The log data bore this out. However users were dissatisfied because what they actually wanted was to be kept informed about when the engineer would turn up, not just "within 2 hours". They also felt it was more important to have technical expertise in the call centre not just message taking and customer care skills.
In that example a survey that asked "Was the call answered quickly enough", "Did the engineer turn up in time", "Was the engineer competent" would score highly. But if you asked "Is the service good" the score would be low. So make sure you're asking the right questions.

There are a number of theoretical models for this sort of feedback. While not that useful on their own they provide a "sanity check" once you have found out your customers specific needs.

2: Get a random sample.

You'll never get truly random samples. There's always some degree of self selection - but if you (say) do a postal survey with only 10% response then it's a very small and specific subset of your users that you're analysing. The statistical validity is low.

Try to force a random sample by making it easy to get the feedback and not allowing the person doing the analysis to (perhaps unwittingly) skew the sample. For example you could call back everyone who has a log number ending in zero who hasn't been surveyed over the last 6 months. It's a simple field to add to the logging database and produces a reasonable sample.

3: Ask measurable questions.

For example ask the subjects to score the particular call between highly satisfied to highly dissatisfied. Don't have an obvious central score - people usually use even numbers so there's no central score.
Also remember you're asking for perceptions not facts. "Was the call answered quickly enough" not "Was the call answered in 3 rings". You're really trying to identify how you're doing against what they consider acceptable and against what they consider to be excellent. It's more important that you're acceptable in all areas rather than excellent in some and unacceptable in others.
I also usually ask then to think about their last encounter rather than try to average things out - there are pros and cons to this.

4: Segment

You will have different user groups. Old/young - callers at different times - business/residential Frequent/occasional users - expert/newbie. Each group has different needs, sometimes conflicting. If you can tune into your segments you will have a definite advantage.


Once you've identified the issues that each customer segment considers as critical there are a number of quite powerful general questions you should keep asking to stay ahead of the game:
* Is the service better or worse than the competition
* What one thing would you improve
* How much would you pay for the service
* Would you recommend to a friend

If you would like to know more please email me. Thanks for dropping in.


I hope you've found this useful. If you did please click on the graphic to vote for this
page as a
Starting Point Hot Site


Aleks de Gromoboy