Forrester Loyalty Metrics
Posted by Jeffrey Henning on Tue, Jun 23, 2009
Often when people talk about “
customer satisfaction”, they really are interested in “customer loyalty”. While
CSAT has been a commonly used customer-satisfaction question for years, different organizations often use quite different loyalty measures, depending on their industry and past experience.
When considering loyalty metrics,
Forrester Research is a useful firm to start with, because Forrester doesn’t provide one overriding loyalty index but instead looks at three questions to determine types of loyalty behavior:
- Willingness to consider the provider for another purchase
- Likelihood to recommend the provider to a friend or colleague
- Reluctance to switch business away from the provider
The actual question text for each is proprietary to Forrester and has not been published. The following are generic versions of these questions.
- Willingness to repurchase: How likely are you to repurchase from us? Not at all likely, Slightly likely, Somewhat likely, Moderately likely, Very likely
- Likelihood to recommend: How likely is it that you would recommend us to a friend or colleague? Not at all likely, Slightly likely, Somewhat likely, Moderately likely, Very likely
- Reluctance to switch: How reluctant are you to switch from us to another provider for these products or services? Not at all reluctant, Slightly reluctant, Somewhat reluctant, Moderately reluctant, Very reluctant
For each measure, Forrester looks at the percent that answer 4 or 5 (“Moderately *” or “Very *” in the scales above). Bruce Temkin has published some
Forrester loyalty results using these measures on his blog.
To speed up the questionnaire for respondents, you can use the same scale as the prior two questions by inverting the last question:
3. How likely are you to switch from us to another provider for these products or services? Not at all likely, Slightly likely, Somewhat likely, Moderately likely, Very likely
If you do that, then for consistency of analysis with the other two measures, reverse the coding of this one question so that “Very likely” is 1 and “Not at all likely” is 5.
These measures are unusual in that Forester elected not to roll them up into a
multi-item scale but kept them separate. Next time around I’ll look at a loyalty index developed by a Vovici partner.