Quantify  
Facebook LinkedIn   Skip Navigation Links
 

Benchmarking comparisons for Employee and Customer Satisfaction Surveys

Why Norms?

When people are looking at the results from their survey, they often wonder how they compare with those found in other organisations. People often think the results don't mean much by themselves, so they want to put them in context by finding out if they are better or worse than other organisations get, or much the same.

What are norms?

To meet this need, we have developed a sophisticated norms management system, comprising a database of over a million data points, each of which represents one informant's  response to one item in a survey. By summarising all these prior responses to the same question, and arranging them in ascending order of the score the informant gave, we can tell you how your result compares with results from those other surveys by reporting the percentage of those other responses which yours is better than, so for each question in your survey which has a parallel in our norms we can report that "Your result was better than X% of the normative data."  

This allows you to put your results in context.

Caution

We have some serious reservations about the use of benchmarking or normative data, however, so while we are pleased to be able to offer a service which meets a wish expressed by many clients, we offer it with the serious health warnings set out below.

Read David Lusty's article Debunking the benchmarking myth which has appeared in a number of journals.

Expectation

This is the most serious drawback of benchmarking, because it can actually lead to a "good" organisation getting worse satisfaction scores than a "bad" one. We think that's perverse, but there's nothing anyone can do to prevent it.

The responses people give to questions about their satisfaction with anything don't depend only on the experience they have had. Exactly the same experience might lead to quite different satisfaction responses depending on what the informant had expected. Someone who had not expected to be treated particularly well might be very pleased if their actual treatment came as a nice surprise, even if it wasn't particularly wonderful. On the other hand, if they had expected excellent treatment, exactly the same not particularly wonderful experience would leave them disappointed so this time they would answer a satisfaction question in a survey less favourably. So satisfaction is a function of experience and expectation. 

The harder you have worked at satisfaction, the higher the expectation of your informants is likely to be and that makes it harder for you to get high satisfaction scores. When you compare your satisfaction scores with those obtained in another organisation, you won't know what their informants were expecting. Another organisation might not in the past have tried half as hard as yours, and their informants, expecting to be very shabbily treated, might have been pleasantly surprised even to be treated in a manner your organisation would regard as unacceptably poor.

Should you then be concerned if your satisfaction scores are no better than that other organisation's? Should you invest management time and scarce resources in dealing with the "problem"? We don't think so.

Of course you won't be comparing your results with just one other organisation's, so the wider norms group will include some "good" and some "bad" organisations. All the same, if your result doesn't compare as well as you had hoped you can never be sure if that is because people are getting a less good experience from your organisation or because they have learned to expect more from you. 

Wording

For their own very good reasons, clients often find a standard item wording inappropriate to their organisation, so they adapt it to meet their needs. Norms include items using the identical standard wording, but they also often include items expressed in equivalent terms. So where the standard item reads I always get the equipment and/or facilities I need to do my job, the data might also include items where clients have preferred I have the equipment and facilities I need to do my job; or We always get the equipment and/or facilities we need to enable us to do our jobs; or I have sufficient resources to enable me to do my job. This pragmatic approach is intended to provide clients with the widest possible normative group to compare with but it doesn't provide like for like comparisons.

Response frame

Clients use different response frames, so while one survey is scored on a four point scale, Very dissatisfied, Dissatisfied, Satisfied, Very satisfied; another might use a seven point agreement scale, Strongly disagree, Disagree, Tend to disagree, Neither agree nor disagree, Tend to agree, Agree, Strongly agree.

To incorporate data gathered using any response frame, each individual response is expressed in the normative database as a percentile score; that is converted as if its response frame had been a scale from 0 to 100. This is a useful device for comparing data gathered using different response frames, but it is a bit rough and ready. For a discussion of the use of percentiles for comparing results on different scales, see Percentiles in the QUANTIFY glossary of terms.

Sequence

The response to a question can be influenced by the other questions which come before it in the questionnaire. Imagine asking people to rate their overall happiness on some scale. If the question is preceded by one asking them to rate the happiness of their marriage / relationship, this narrows the perception of the overall question, so people reply to it much more in the context of their relationship than they would if the general question was put before the marriage / relationship one.

Even if we compare data derived from items using identical text the different questions which have preceded it in the various questionnaires will have coloured responses in varying ways which we can't predict or control, so once again, we aren't strictly comparing like with like.

Conclusion

All these influences mean that while normative comparisons might give some clue to how your survey results relate to other people's, they should be approached with great caution and corroborative evidence should be sought before you invest effort and resources in addressing any perceived competitive disadvantage.

Team

More information

Debunking the Benchmarking Myth
David Lusty's article has appeared in several journals.
More articles and literature from Quantify

Web Surveys

Visit our sample web survey

If this panel gets in your way, click here  to reduce or restore its width, or here to remove it for the duration of this visit.