Welcome to the Listening Post!

Inspired by the folk Apostle Model, TNS has its own customer-loyalty segmentation, a quadrant analysis contrasting likelihood to recommend and likelihood to repurchase.

 

 

 

Note: I swapped the axes from the customary presentation that TNS uses to make the segmentation more clearly match up with the Apostle Model.

 

The four segments of this model:

• Champion - high repurchase intentions, high likelihood to recommend – true apostles of your brand, who practice what they preach
• Moral Supporter – low repurchase intentions, high likelihood to recommend – false prophets of your brand
• Rebel - low repurchase intentions, low likelihood to recommend – these apostates won’t be customers long
• Captive - high repurchase intentions, low likelihood to recommend – these hostages can’t switch because of corporate policy or financial considerations

The actual questions, taken from the TNS CLI template:

• Likelihood to recommend - "Based on your experience, how likely would you be to recommend [organization/brand] to a friend or colleague looking for [product/service category]?  Definitely would, Probably would, Might or might not, Probably would not, Definitely would not"
• Likelihood to repurchase – "For your next similar purchase of [product/service category], how likely would to be to buy  from [organization/brand] again?  Definitely would, Probably would, Might or might not, Probably would not, Definitely would not"

The TNS loyalty model is most well-known to Microsoft partners, for whom TNS will conduct this customer-loyalty segmentation for free (e.g., at Microsoft’s expense). The segmentation seems to suffer from not segmenting enough. In segmentations made public by two Microsoft partners, each showed samples sizes of two dozen respondents, with 100% champions; the partner average itself currently shows 88.6% of customers are champions.

 

This model is great in its focus on loyalty behaviors, and it is certainly better than the traditional Apostle Model in aligning its segmentations with actual behavior: the Apostle Model itself segments a group into “Loyalists/Apostles” but doesn’t have any measure of likelihood to recommend, a key behavior that would be expected of a group that is evangelizing a particular vendor. (Though Bob E. Hayes, among others, shows how CSAT can drive advocacy loyalty.)

TNS uses a five-point bipolar scale instead of a five-point unipolar scale, which will have greater reliability and validity. Accordingly, I suggest using this scale instead: Not at all likely, Slightly likely, Moderately likely, Very likely, Completely likely.

 

A segmentation that classifies almost 90% of your customers as champions isn’t going to inspire your organization to improve.  In truth, you want customers who are completely likely to repurchase and recommend:  only classify the “completely likely” as champions.  Your customer satisfaction initiatives should be designed around driving that type of behavior. Instead of using a traditional quadrant analysis, I refactor the analysis to be analogous to the Apostle Model: